United States v. Phillips

Full title: United States of America, Plaintiff-Appellee, v. Trinity Phillips…

Court: United States Court of Appeals, Sixth Circuit

Date published: Nov 28, 2022

54 F.4th 374 (6th Cir. 2022)

Opinion

No. 21-5762

11-28-2022

UNITED STATES of America, Plaintiff-Appellee, v. Trinity PHILLIPS, Defendant-Appellant.

ARGUED: Michael C. Holley, OFFICE OF THE FEDERAL PUBLIC DEFENDER, Nashville, Tennessee, for Appellant. Joshua K. Handell, UNITED STATES DEPARTMENT OF JUSTICE, Washington, D.C., for Appellee. ON BRIEF: Michael C. Holley, Ronald C. Small, OFFICE OF THE FEDERAL PUBLIC DEFENDER, Nashville, Tennessee, for Appellant. John M. Pellettieri, UNITED STATES DEPARTMENT OF JUSTICE, Washington, D.C., S. Carran Daughtrey, UNITED STATES ATTORNEY'S OFFICE, Nashville, Tennessee, for Appellee.

BOGGS, Circuit Judge.

ARGUED: Michael C. Holley, OFFICE OF THE FEDERAL PUBLIC DEFENDER, Nashville, Tennessee, for Appellant. Joshua K. Handell, UNITED STATES DEPARTMENT OF JUSTICE, Washington, D.C., for Appellee. ON BRIEF: Michael C. Holley, Ronald C. Small, OFFICE OF THE FEDERAL PUBLIC DEFENDER, Nashville, Tennessee, for Appellant. John M. Pellettieri, UNITED STATES DEPARTMENT OF JUSTICE, Washington, D.C., S. Carran Daughtrey, UNITED STATES ATTORNEY'S OFFICE, Nashville, Tennessee, for Appellee.

Before: BOGGS, LARSEN, and DAVIS Circuit Judges.

BOGGS, J., delivered the opinion of the court, in which DAVIS, J., joined. LARSEN, J. (pp. –––– – ––––), delivered a separate opinion concurring in the judgment only.

BOGGS, Circuit Judge.

When it passed the PROTECT Act in 2003, Congress required the United States Sentencing Commission to vary penalties for child-pornography offenses depending on the number of images involved. The Commission accordingly implemented that method of calculating penalties in the Sentencing Guidelines. ¹ Addressing what it perceived to be an ambiguity in Congress's command, the Commission added an application note in the Guidelines commentary instructing courts to equate one video to seventy-five images when calculating the applicable Guidelines sentencing range.

1 The significance of Guidelines commentary is discussed in Section II.A, infra .

For almost twenty years, courts have relied on this "75:1 Rule" when sentencing defendants convicted of possessing videos containing child pornography. Recent Supreme Court precedent, however, has clarified when courts can defer to an agency's interpretation of its regulations (by applying so-called Auer deference). Defendant-Appellant Trinity Phillips argues that this recent clarification means that a sentencing court can no longer rely on the 75:1 Rule, and that the court erred in relying on it when imposing his sentence. We disagree and affirm that sentence as imposed by the district court.

I. BACKGROUND

Defendant-Appellant Trinity Phillips was convicted of possessing child pornography. Law enforcement became aware of Phillips during the fallout of his romantic relationship with Samantha Melford. The pair had met online and conducted a long-distance affair—both were married. Their relationship was characterized by violence, sexual objectification, and a "master/slave" dynamic requiring Melford to submit to Phillips's will. During the course of the relationship, Melford filmed herself performing sexual and pseudo-sexual acts on or near her 5-year-old half-sister. She then sent those videos to Phillips.

Melford eventually moved in with Phillips and his wife and children, but left following an argument in 2018. She then informed police about Phillips's interest in and possession of child pornography. Investigators obtained a warrant and searched his residence in Tennessee. The search revealed multiple laptops and other electronic devices. One of those laptops contained backup files from Phillips's cell phone, including a file entitled "Note 5." That folder contained two subfolders: "Sister 2" and "Kik." "Sister 2" contained 82 videos and 3 still images of child pornography, all of which had been produced by Melford. "Kik" contained 9 other videos that also depicted child pornography. And forensic investigators found 169 thumbnail images that were indicative of previously viewed or deleted child pornography. In total, therefore, law enforcement obtained 172 still images and 91 videos.

Phillips was charged with one count of knowingly receiving child pornography, in violation of 18 U.S.C. §§ 2552A(a)(2)(A) and 2552A(b), and one count of knowingly possessing child pornography, in violation of 18 U.S.C. § 2552A(5)(B) and 2552A(b). He pleaded guilty as charged.

The Guidelines recommend applying different sentence enhancements depending on the number of "images" involved in the offense. Specifically, the Guidelines include an "image table," which lays out different enhancement levels for different numbers of images:

If the offense involved—

"(A) at least 10 images, but fewer than 150, increase by 2 levels;

"(B) at least 150 images, but fewer than 300, increase by 3 levels;

"(C) at least 300 images, but fewer than 600, increase by 4 levels; and

"(D) 600 or more images, increase by 5 levels."

U.S.S.G. § 2G2.2(b)(7).

An application note in the commentary to Guideline § 2G2.2 instructs how to determine the number of images involved in the offense. U.S.S.G. § 2G2.2(b)(7) n.6(B). According to that note, each video counts as 75 images, and if the length of the video is substantially more than five minutes, an upward departure may be warranted. Ibid. This conversion ratio is referred to as the "75:1 Rule."

In calculating the appropriate sentence range for Phillips pursuant to the image table, the probation office calculated that, pursuant to the commentary to application note 6, his offense involved 6,997 images—172 still images and 75 images for each of the 91 videos. That number of images led to a five-level enhancement. Phillips objected to that calculation, arguing that each video should count as one image, not 75, and that, consequently, he should only be responsible for 263 images and thus only a corresponding three-level enhancement.

The district court held an evidentiary hearing to address this and other issues. At the end of that hearing, the district court found that the five-level enhancement was appropriate in light of the 75:1 Rule. ²

2 The district judge also seemed to suggest that he would have concluded more than 600 images were involved even without applying the 75:1 Rule. During the hearing, he stated that:

This enhancement ... is appropriate. Even ... applying the evidence presented at the hearing, the number of images would be more than 600. And, finally, applying the rationale and argument set forth in the presentence report, which is a very, very conservative 15 second per frame [sic], we would still have more than 600 images. So, by any of those standards, the five-level enhancement applies.

At sentencing, the district court applied the five-level enhancement. With that enhancement, Phillips's Guidelines range was 121 to 151 months of imprisonment. If the district court had instead applied the three-level enhancement requested by Phillips, his range would have been 97 to 121 months of imprisonment. He was ultimately sentenced to 151 months of imprisonment. This appeal followed.

II. ANALYSIS

Phillips argues that the district court was wrong to rely on the 75:1 Rule. He suggests that recent Supreme Court precedent has rendered reliance on that application note impermissible. The Government disagrees and argues, in the alternative, that Phillips's sentence should be affirmed even if reference to the 75:1 Rule is discarded. We hold that the district court did not err in relying on the 75:1 Rule.

A. History of the 75:1 Rule

The Guidelines originated with the Sentencing Reform Act of 1984, when Congress tasked the Commission with creating sentencing ranges for various offenses. Stinson v. United States , 508 U.S. 36, 40–41, 113 S.Ct. 1913, 123 L.Ed.2d 598 (1993). While the Commission is authorized to amend the Guidelines on its own, it must allow Congress six months to review any amendments, and also to allow any amendments to go through a period of notice and comment. 28 U.S.C. § 994(p), (x). While sentencing courts have discretion to deviate from the Guidelines recommendation, that discretion is somewhat limited. United States v. Havis , 927 F.3d 382, 385 (6th Cir. 2019) (en banc) (per curiam) (citing Peugh v. United States , 569 U.S. 530, 543, 133 S.Ct. 2072, 186 L.Ed.2d 84 (2013) ). A sentencing judge "cannot stray from a defendant's Guidelines range, for example, without first giving an adequate explanation." Ibid.

In addition to the Guidelines themselves, the Commission has included "application notes" in "commentary" intended to interpret, explain, or otherwise aid the sentencing court in choosing a sentence. See United States v. Riccardi , 989 F.3d 476, 484 (6th Cir. 2021) (citing U.S.S.G. § 2B1.1 cmt. nn.1–8). This "commentary," however, is not subject to the same procedural safeguards as the Guidelines themselves. The Commission may amend the commentary unilaterally, without notice and comment or congressional review. Havis , 927 F.3d at 386. While some courts initially doubted whether sentencing courts were bound by the commentary's interpretation of the Guidelines, the Supreme Court ultimately held that sentencing courts owe the commentary the same kind of deference owed to an agency's interpretation of its own regulations in other contexts. Stinson , 508 U.S. at 45, 113 S.Ct. 1913.

The commentary at issue here came about in response to the PROTECT Act of 2003, Pub. L. No. 108-21, 117 Stat. 650. Within that statute, Congress took the unusual step of amending the Guidelines directly. See United States v. McNerney , 636 F.3d 772, 777 (6th Cir. 2011). It required the Commission to adopt the image table. The purpose of the image table was to increase penalties "based on the amount of child pornography involved in the offense." H.R. Rep. No. 108-66 at 59. The Commission duly implemented the image table as decreed by Congress, and it is included verbatim in Guideline § 2G2.2(b)(7).

Congress did not, however, define "image." Nor did it explain how to determine the number of images in a video. To combat this uncertainty, the Commission sought public comment on whether it was necessary to include instructions on "counting images" and, if so, how to determine the number of images in a video. 68 Fed. Reg. 75340, 75353 (Dec. 30, 2003). The Commission posed a hypothetical example: "if a video includes numerous scenes, each of which portrays the same minor engaging in sexually explicit conduct with a different adult, is each scene with a different adult to be considered a separate image?" Ibid .

Following this solicitation of public input, the Commission adopted the 75:1 Rule and other application notes. According to the Commission, it

ultimately determined that because each video contained multiple images [a video] should be counted as more than one image. Given that the image table enacted by Congress assigned a 2-level increase for between ten images and 150 images, and a 3-level increase for 150 to 300 images, the Commission adopted a definition of video that considered each video to contain 75 images, squarely in the middle of the 2-level increase range. The Commission also expressly authorized an upward departure if the video was substantially longer than five minutes.

History of the Child Pornography Guidelines , Oct. 2009, at 43–44, available at https://www.ussc.gov/sites/default/files/pdf/research-and-publications/research-projects-andsurveys/sex-offenses/20091030_History_Child_Pornography_Guidelines.pdf (last visited Oct. 11, 2022).

As it stands, application note 6 reads:

Application of Subsection (b)(7).—

(A) Definition of "Images."—"Images" means any visual depiction, as defined in 18 U.S.C. § 2256(5), that constitutes child

pornography, as defined in 18 U.S.C. § 2256(8).

(B) Determining the Number of Images.—For purposes of determining the number of images under subsection (b)(7) :

(i) Each photograph, picture, computer or computer-generated image, or any similar visual depiction shall be considered to be one image. If the number of images substantially underrepresents the number of minors depicted, an upward departure may be warranted.

(ii) Each video, video-clip, movie, or similar visual depiction shall be considered to have 75 images . If the length of the visual depiction is substantially more than 5 minutes, an upward departure may be warranted.

U.S.S.G. § 2G2.2 cmt. n.6 (emphasis added).

And the term "visual depiction" is defined by statute, which explains that the term

includes undeveloped film and videotape, data stored on computer disk or by electronic means which is capable of conversion into a visual image, and data which is capable of conversion into a visual image that has been transmitted by any means, whether or not stored in a permanent format.

18 U.S.C. § 2256(5).

With this background, we turn to Phillips's assertion that the 75:1 Rule must be cast aside.

B. Whether the Guidelines Commentary is Entitled to Auer Deference

1. The Kisor Standard

Phillips argues that the Supreme Court's recent decision in Kisor v. Wilkie , ––– U.S. ––––, 139 S. Ct. 2400, 204 L.Ed.2d 841 (2019), has altered the level of deference a sentencing court owes to the Guidelines commentary. He claims that the 75:1 Rule does not meet Kisor 's new standard and that the district court therefore erred in relying on that application note. Instead, he claims that the district court should have applied a definition of "image" that counts each video as a single image when calculating his Guidelines range.

To start, we need not decide whether Kisor controls here because we have already held that it does. Prior to Kisor , our analysis of the Guidelines commentary was guided by Stinson . That case instructed us to apply what was then called Seminole Rock deference (but is now called Auer deference) to commentary unless the commentary's interpretation was "plainly erroneous or inconsistent" with the guideline itself. Stinson , 508 U.S. at 38, 113 S.Ct. 1913 ; see Bowles v. Seminole Rock & Sand Co. , 325 U.S. 410, 414, 65 S.Ct. 1215, 89 L.Ed. 1700 (1945), Auer v. Robbins , 519 U.S. 452, 461, 117 S.Ct. 905, 137 L.Ed.2d 79 (1997). But we recently observed that Kisor "clarified Auer 's narrow scope" and provided the framework that we must follow in determining whether to defer to the Guidelines commentary. Riccardi , 989 F.3d at 484–85. We therefore look to Kisor to see whether a sentencing court may rely on the 75:1 Rule.

Kisor concerned whether Auer deference should remain a factor when interpreting agency regulations. The motivating principle of Auer deference is "rooted in a presumption about congressional intent—a presumption that Congress would generally want the agency to play the primary role in resolving regulatory ambiguities." 139 S. Ct. at 2412. This presumption is partly due to a feeling that the agency that developed a rule is in a better position to understand what it means, but also "stems from the awareness that resolving genuine regulatory ambiguities often ‘entail[s] the exercise of judgment grounded in policy concerns.’ " Id. at 2413 (quoting Thomas Jefferson Univ. v. Shalala , 512 U.S. 504, 512, 114 S.Ct. 2381, 129 L.Ed.2d 405 (1994) ).

Still, a court cannot reflexively defer to an agency's interpretation. "First and foremost, a court should not afford Auer deference unless the regulation is genuinely ambiguous." Id. at 2415. We cannot defer to the agency "if there is only one reasonable construction of a regulation." Ibid. "And before concluding that a rule is genuinely ambiguous, a court must exhaust all the ‘traditional tools’ of construction." Ibid. (quoting Chevron U.S.A., Inc. v. Nat. Res. Def. Council, Inc. , 467 U.S. 837, 843 n.9, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984) ). That strategy requires the court to "carefully consider[ ] the text, structure, history, and purpose of a regulation." Ibid. (quotation marks omitted).

Concluding that a rule is "genuinely ambiguous," though, is only step one. "[T]he agency's reading must still be ‘reasonable’ " and "come within the zone of ambiguity the court has identified after employing all its interpretive tools." Id. at 2415–16 (quoting Thomas Jefferson , 512 U.S. at 515, 114 S.Ct. 2381 ). "The text, structure, history, and so forth at least establish the outer bounds of permissible interpretation." Id. at 2416.

But even for "reasonable" interpretations, "a court must make an independent inquiry into whether the character and context of the agency interpretation entitles it to controlling weight." Ibid. While there is no "exhaustive test" to determine whether that is the case, the Court has "laid out some especially important markers for identifying when Auer deference is and is not appropriate." Ibid. First, the interpretation must be the official position of the agency. Ibid. "Next, the agency's interpretation must in some way implicate its substantive expertise." Id. at 2417. "Finally, an agency's reading of a rule must reflect ‘fair and considered judgment.’ " Ibid. (quoting Christopher v. SmithKline Beecham Corp. , 567 U.S. 142, 155, 132 S.Ct. 2156, 183 L.Ed.2d 153 (2012) ).

2. Whether the Guideline is Genuinely Ambiguous

We must first determine, then, whether a genuine ambiguity exists. Phillips frames the question as whether the term "image" is itself ambiguous. That is accurate in some sense, in that it is impossible to determine how many images are in a video without understanding the term "image." But the ambiguity is wider than that; the questions are "how many images are in a video?" and "how do videos map on to the image table?"

The Government suggests that this decision has already been made for us by United States v. Geerken , 506 F.3d 461 (6th Cir. 2007). There, we had to decide whether including the 75:1 Rule in the commentary was "clarifying" or "substantive." Id. at 465. If substantive, then it could not be applied retroactively to defendants at sentencing; if clarifying, then it "may be applied retroactively to discern the Sentencing Commission's intent regarding the application of a pre-amendment guideline." Ibid. (citing United States v. DeCarlo , 434 F.3d 447, 458–59 (6th Cir. 2006) ).

We concluded that the failure to define "image" left an ambiguity in the 2003 amendment to the Guidelines, and that the 75:1 Rule was intended to aid the sentencing court in determining the number of images in a video for image-table purposes. Id. at 466. But that analysis does not automatically carry over to this case. Our task in Geerken was primarily to determine whether the Commission itself was seeking to clarify application of the image table. It is true that Geerken held there was an ambiguity in reconciling videos with the image table. But that case was decided before Kisor , which instructed us to apply the traditional tools of statutory construction in deciding whether genuine ambiguity exists. Geerken did not, therefore, hold that the image table reflects the kind of "genuine ambiguity" that must exist to be afforded Auer deference. The Geerken court did not need to engage in that kind of exercise to resolve the issue before it. So, while we need not disturb the holding in Geerken , we cannot rely on it directly to conclude that the image table warrants Auer deference in the wake of Kisor .

Applying that analysis, we first address Phillips's argument that the image table is not ambiguous. Indeed, he suggests that there is only one plausible meaning: Every video counts as one image. The argument goes that when Congress included the image table in the PROTECT Act, existing child-pornography statutes treated a "visual image" and a "visual depiction" the same. Phillips refers to 18 U.S.C. § 2256, which includes two relevant definitions. The first defines "visual depiction" as

[i]nclud[ing] undeveloped film and videotape , data stored on computer disk or by electronic means which is capable of conversion into a visual image, and data which is capable of conversion into a visual image that has been transmitted by any means, whether or not stored in a permanent format.

18 U.S.C. § 2256(5) (emphasis added). The second defines "child pornography" as:

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

(A) the production of such visual depiction involves the use of a minor engaging in sexually explicit conduct;

(B) such visual depiction is a digital image, computer image, or computer-generated image that is, or is indistinguishable from, that of a minor engaging in sexually explicit conduct; or

(C) such visual depiction has been created, adapted, or modified to appear that an identifiable minor is engaging in sexually explicit conduct.

Id. § 2256(8).

Phillips also points to the fact that the commentary to Guideline § 2G2.6(b)(7) defines an image as "any visual depiction, as defined in 18 U.S.C. § 2256(5), that constitutes child pornography, as defined in 18 U.S.C. § 2256(8)." U.S.S.G. § 2G2.2(b)(7) cmt. n.6(A). To Phillips, this means that the Commission itself adopted the preexisting definition of visual depiction.

We do not agree that the definition of "visual depiction" equates that term with "visual image." Instead, the definition of "visual depiction" broadens the reach of the term to include various items that might not otherwise normally be considered a "visual depiction," such as undeveloped film, or computer data that is capable of being converted into a visual image. And while the definition of "child pornography" does explain that videos are a type of visual depiction, it provides no indication that each individual visual depiction is equivalent to one "image" as that term is used in the Guideline.

Even assuming, arguendo , that Phillips's position that one video equals one image is a reasonable interpretation, it is not the only reasonable interpretation. We must go on, then, to apply the tools of construction to determine whether genuine ambiguity exists.

We start with the text of the Guideline itself. United States v. Hill , 963 F.3d 528, 532 (6th Cir. 2020). The term "image," of course, means different things in different contexts. The most relevant definition of "image" is a "physical or digital representation of something, originally captured using a camera from visible light, and typically reproduced on paper, displayed on screen, or stored as a computer file." Image , Oxford English Dictionary (2022). And a "video" is, most relevantly, "a recording of moving visual images ... in a digital format." Video , Oxford English Dictionary (2022). So it is clear that a "video" is comprised of some number of images. But how many?

One possible answer is that the number of images in a video is equivalent to the number of frames. To get to this answer, one might consider that videos are akin to films, and films traditionally are comprised of a series of still images that are displayed rapidly to create the illusion of continuous movement. See Frame , Oxford English Dictionary (2022) ("[O]ne of the individual images on a strip of film; (later also) a single complete image in a series forming a television picture, film, or video sequence."). A film, then, could be considered nothing but a collection of singular images. As the provocative French filmmaker Jean-Luc Godard said: "Photography is truth. The cinema is truth 24 times per second." The Yale Book of Quotations 313 (Fred R. Shapiro ed., 2006).

But even in the cinematic context, "image" has multiple meanings. When one says that "the film contained a number of powerful images," that does not refer to frames. One can imagine asking a friend if they enjoyed a recent showing of Ingmar Bergman's The Seventh Seal at the local arthouse cinema. Perhaps the friend did enjoy it. If the friend were asked if the film contained any striking "images," they might respond "Yes, I found the iconic image of Death playing chess with the crusader on a deserted beach very memorable." If one followed up and asked which frame in particular the friend was referring to, they would respond with a puzzled expression. Mostly because the human eye cannot perceive any individual frame, but more so because the "image" of chess-playing Death does not correspond to any particular frame, nor to any singular still moment at all.

Equating the number of frames in a video with the number of images, therefore, is not necessarily the only way to understand the image table even as a textual matter. Indeed, there likely are any number of frames in a given video that do not include the prohibited subject matter at all, as when the camera scans a room for setting or settles on other innocuous items for instance. It would seem contrary to congressional intent to include such frames in the count of images that lead to increased punishment. And parsing out such frames would require courts to undertake a frame-by-frame analysis of each video to ascertain the number of frames that include illegal images, an onerous and unrealistic task given the multitude of frames in any one video and the many cases that involve multiple videos. The frame-as-image analysis takes no account of this complication and pointing out this and other complications with the dictionary-definition analysis in no way suggests that the majority here "seeks to protect" any particular class of defendants. As we recently observed in United States v. Tate regarding textual interpretation, "literal or dictionary definitions of words will often fail to account for settled nuances or background conventions that qualify the literal meaning of language and, in particular, of legal language." 999 F.3d 374 (6th Cir. 2021) (internal citation omitted); see also Antonin Scalia, A Matter of Interpretation 23 (1997) ("A text should not be construed strictly, and it should not be construed leniently; it should be construed reasonably, to contain all that it fairly means."). Hence, we commonly look at the broader context in which the term at issue has been used and have expressly done so in interpreting the words of the Sentencing Commission. See Stinson , 508 U.S. at 44, 113 S.Ct. 1913. We are not, thus, limited to looking at the text alone. We also "must carefully consider" the structure, history, and purpose of the Guideline as well. Kisor , 139 S. Ct. at 2415 (quotation marks omitted) (emphasis added). The structure, history, and purpose also lead us to conclude that the Guideline is ambiguous.

The purpose of the image table is not in doubt; it was intended to increase the penalty by an amount corresponding to the number of images involved in the offense. And it did so by increasing the penalty when the number of images reached specific numerical thresholds, capping out at 600. If, for instance, "images" meant "frames," then possessing any video would nearly automatically vault the offender to the top of the range, thereby obviating the purpose of prescribing different levels. For example, assuming a frame rate of 24 flames per second, the penalties would correspond to videos of the following lengths:

Offense involves:	Enhancement	Video length (in seconds)
At least 10 images, but fewer than 150	2-level increase	.42-6.25
At least 150 images, but fewer than 300	3-level increase	6.25-12.5
At least 300 images, but fewer than 600	4-level increase	12.5-25
600 or more images	5-level increase	25+

If a defendant possessed one video, the video would need to be just 25 seconds long to warrant the maximum increased penalty. And differences of only a few seconds lead to vastly different penalties. Construing "image" to mean "frame" would mean that videos are disproportionately counted as compared to still images. Not only is there no indication from Congress that the intent of the image table was to penalize videos to that degree, but the image table's precision indicates that Congress sought to vary the penalty based on proportionate quantities of images, not quarter-second differences.

Moreover, defining "images" to mean "frames" is utenable in that it would lead to decidedly unreasonable results. For example, imagine an offender who had 100 photographs, each depicting a different victim. Contrast that with an offender who has a single 30-second video depicting a single victim. Under the construction where an "image" equals a "frame," the former offender would receive a one-level increase while the latter would receive a five-level increase. Yet that result is totally unmoored from the goal of differentiating penalties based on the amount of child pornography.

Another example is a slow-motion video. Some super-slow-motion videos can reach 1000 frames per second. Imagine that one offender films a five-second scene with a regular camera, while another offender films that same scene with a slow-motion camera. Again, the former would receive the minimum penalty while the latter would receive the maximum; despite the actual conduct and content being the same. Or imagine a video gently panning over a photograph, as in the documentaries of Ken Burns. A ten-second video of that image would warrant a three-level increase, while that same image, on its own, would not warrant even a one-level enhancement.

The term "image," therefore, when used in the context of the image table, is ambiguous when it comes to determining the number of images in a video. We conclude that after applying the traditional tools of statutory construction, determining the number of images in a video is the kind of genuinely ambiguous exercise that the Commission was entitled to address in commentary.

C. Whether the 75:1 Rule is Within the Zone of Ambiguity

We then turn to the second step of the Kisor inquiry to determine whether the Commission's construction is within the "zone of ambiguity." On this question, Kisor instructs us to look to the text, structure, and history of the Guideline to see if the 75:1 Rule is a reasonable way to interpret the image table as applied to videos. Soon after the PROTECT Act, the Commission sought input on how to define "images" in the video context. As the Commission itself explains, it

ultimately determined that because each video contained multiple images it should be counted as more than one image. Given that the image table enacted by Congress assigned a 2-level increase for between ten images and 150 images, and a 3-level increase for 150-300 images, the Commission adopted a definition of video that considered each video to contain 75 images, squarely in the middle of the 2-level increase range. The Commission also expressly authorized an upward departure if the video was substantially longer than five minutes.

History of the Child Pornography Guidelines at 43–44.

In doing so, the Commission acknowledged the disproportionate results that would occur if it adopted either Phillips's one-video-one-image calculation or a one-frame-one-image calculation. Given the context of the image table, the Commission adopted a rule that would ensure that Congress's precise delineation would be meaningful. The 75:1 Rule takes into account the text, acknowledging that a video is comprised of multiple images, as both terms are commonly understood. It also takes into account the history and structure of the image table; that is, the scale of the differentiation—10 to 600—such that Congress's choice of those figures would have real-world consequences.

Phillips claims that, even assuming the image table is ambiguous, the 75:1 Rule is outside of the zone of ambiguity. In doing so, he cites our recent Riccardi decision. That case also considered whether a sentencing court could apply Auer deference to Guidelines commentary. The defendant there had been convicted of stealing gift cards. 989 F.3d at 479. The relevant Guideline instructed the sentencing court "to ‘increase the offense level’ in incremental amounts based on the amount of the ‘loss’ (measured in dollars)." Id. at 486 (quoting U.S.S.G. § 2B1.1(b)(1) ). Like here, the critical term "loss" was undefined in the Guideline. Id. at 481. Instead, commentary to the Guideline instructed that "loss" as applied to gift cards "shall not be less than $500" per card. Id. at 482 (quoting U.S.S.G. § 2B1.1 cmt. n.3(F)(i)). The defendant argued that this $500 figure was not entitled to deference. Id. at 483.

We agreed and held that "a $500 mandatory minimum cannot be described as an interpretation of the word loss." Ibid. Putting aside the question of whether the term "loss" was genuinely ambiguous, we held that the choice of a $500 amount was not within the zone of ambiguity. Id. at 486. "No reasonable person would define the ‘loss’ from a stolen gift card as an automatic $500. Rather, the ‘amount’ of the loss or ‘damage’ to the victim from a gift-card theft in any case will turn on such fact-dependent things as the value of the gift card or the costs of replacing it." Ibid. The disconnect between the $500 figure and the empirically measurable amount of the loss was especially striking in the defendant's case since the average value of the stolen gift cards was only $35 each. Id. at 480.

Riccardi means that the Commission is not interpreting its Guidelines when it adopts an unreasonable numerical formula just because the loss may be difficult to quantify. See id. at 487. We do not, however, understand Riccardi to mean that the Commission cannot be interpreting within the zone of ambiguity any time it uses a numerical approximation. The interpretive question here is different than in Riccardi . The "loss" represented by a stolen gift card is, in theory, a figure that the sentencing court could ascertain as a factual matter. The number of "images" in a video, however, is not a question that can be answered without making some sort of interpretation. As discussed above, the word "image" in the image-table context does not have only one reasonable meaning. The 75:1 figure was not based on a compromise due to the empirical challenge of determining the number of frames in a video. Instead, facing this interpretive challenge, informed by the history and structure of the image table, the Commission made a choice of 75:1.

We do not hold that the Commission could have picked any number out of a hat. Instead, we hold only that the 75:1 Rule fits within the zone of ambiguity because it considers (1) the fact that videos contain multiple images; (2) the image table's purpose of tying offense levels to the number of images; and (3) Congress's choice to create four different tiers of punishment, based on a scale of 10 to 600 images. The 75:1 Rule is not based on an independent policy choice by the Commission to implement some other sentencing or adjudicative goal. Instead, the Commission chose that ratio in direct response to the challenges of applying the image table laid out by Congress. For that reason, it is a reasonable interpretation of the term "images" as applied to videos.

D. Whether the Character and Context of the 75:1 Rule Suggest Controlling Weight

Finally, Kisor requires us to consider whether the "character and context of the agency interpretation entitles it to controlling weight." Kisor , 139 S. Ct. at 2416. Looking at the three touchstones laid out in Kisor , we conclude that it does. First, the Guidelines commentary represents the official position of the Commission—this is not an ad hoc pronouncement. See Kisor , 139 S. Ct. at 2416. Indeed, the Commission solicited public comments on applying the image table. Second, interpreting the Guidelines to promote sentencing goals squarely implicates the Commission's "substantive expertise." See id. at 2417. "The functional purpose of the commentary ... is to assist in the interpretation and application [of the Guidelines,] which are within the Commission's particular area of concern and expertise and which the Commission itself has the first responsibility to formulate and announce." Stinson , 508 U.S. at 45, 113 S.Ct. 1913. Third, the interpretation was not instituted for any post hoc purpose, but was put directly into the commentary more than fifteen years ago following a period of public consultation, and therefore reflects the Commission's "fair and considered judgment." See Kisor , 139 S. Ct. at 2417 ("[A] court should decline to defer to a merely convenient litigating position or post hoc rationalization[n] advanced to defend past agency action against attack.") (quotation marks omitted).

Phillips argues that the Commission's interpretation of the Guideline has no weight because the image table was initially drafted and instituted by Congress, and the Guideline merely restates a statutory term. He cites Federal Express Corporation v. Holowecki for the proposition that we cannot grant Auer deference where the term was "not a construct of the agency's regulations" but instead "a term Congress used in the underlying statute that has been incorporated in the regulations by the agency." 552 U.S. 389, 398, 128 S.Ct. 1147, 170 L.Ed.2d 10 (2008). Yet the Court did not hold that Auer deference can never be applied in that kind of situation. It noted only that it "could be argued" that Auer deference would not apply, and also stated that "[i]t is not necessary to hold" whether Auer deference applied. Id. at 399, 128 S.Ct. 1147. That case was, moreover, decided prior to Kisor , which laid out a more specific blueprint for determining whether we should grant Auer deference to the Guidelines commentary.

The Government also distinguishes the Guidelines commentary from the interpretation at issue in Holowecki by pointing out that, in the PROTECT Act, Congress specifically authorized the Commission to "make further amendments to the sentencing Guidelines, policy statements, or official commentary." PROTECT Act § 401(j)(3). This, coupled with the fact that the Commission's interpretation of the Guidelines contains all three of the factors that Kisor outlines as supportive of deference, confirms for us that the 75:1 Rule has the character and context of an interpretation entitled to controlling weight.

III. CONCLUSION

The 75:1 Rule, therefore, is the kind of agency interpretation that warrants Auer deference in the wake of Kisor . The sentencing court did not err in relying on it when determining Phillips's sentence.

For the foregoing reasons, we AFFIRM the sentence.

CONCURRING IN THE JUDGMENT

LARSEN, Circuit Judge, concurring in the judgment only.

How is a court to respond when the question before it involves the interpretation of an agency rule? Over decades, we lower courts developed a habit of deferring reflexively to the agency's interpretation under Seminole Rock and Auer , rather than first tackling the interpretative question ourselves, to see whether the rule was "genuinely ambiguous." Kisor v. Wilkie , ––– U.S. ––––, 139 S. Ct 2400, 2414, 204 L.Ed.2d 841 (2019). Three years ago, the Supreme Court told us to stop. The Court did not mince words: "[T]he possibility of deference can arise only if a regulation is genuinely ambiguous. And when we use that term, we mean it—genuinely ambiguous, even after a court has resorted to all the standard tools of interpretation." Id. In United States v. Riccardi , this court confirmed that Kisor 's admonition applied to the Sentencing Guidelines too. 989 F.3d 476, 486 (6th Cir. 2021). These were important decisions. They reminded us that judges have a duty to interpret the law, even when administrative agencies are involved. But old habits are hard to break. Today's decision is proof. No fair reading of Kisor and Riccardi would permit us to defer to the Sentencing Commission's conclusion that the word "image" means 1/75th of every video. Nevertheless, the majority opinion rolls right through Kisor 's stop sign, reflexively deferring to an agency's non-interpretation of an unambiguous Sentencing Guideline. So I concur in the judgment only.

* * *

This case asks us to interpret the term "image" as it is used in U.S.S.G. § 2G2.2(b)(7), which sets increasing sentence enhancements based on the number of images involved in a child-pornography offense. We must determine how many images were contained in the 91 child-pornography videos Trinity Phillips possessed. Four meanings of "image" are on the table: "one video" (Phillips's interpretation); "1/75th of a video" (the Commission's rule); "frame" (the government's fallback position); and the majority opinion's amorphous definition drawn from two friends' hypothetical viewing of an Ingmar Bergman film (more on that later).

The majority concludes that we must defer to the rule found in the Sentencing Commission's commentary, which provides that "[e]ach video ... shall be considered to have 75 images." But Kisor and Riccardi are crystal clear that at least two conditions must be met before a court may defer to an agency's interpretation of its own rule. The rule must be "genuinely ambiguous." Riccardi , 989 F.3d at 486 (quoting Kisor , 139 S. Ct. at 2415 ). And the agency's interpretation must be " ‘within the zone of ambiguity’ that exists." Id. at 480 (quoting Kisor , 139 S. Ct. at 2400 ). The government's plea for deference, which the majority readily indulges, ticks neither box.

I.

I begin by explaining why the Sentencing Commission's 75:1 rule cannot be entitled to deference. A court may not defer to an "agency's reading" of its own regulation unless that reading "fall[s] ‘within the bounds of reasonable interpretation.’ " Kisor , 139 S. Ct. at 2416 (quoting City of Arlington v. FCC , 569 U.S. 290, 296, 133 S.Ct. 1863, 185 L.Ed.2d 941 (2013) ). "In other words, it must come within [the regulation's] zone of ambiguity." Id. at 2415–16. The bounds of that zone are set by the court's interpretation of the rule—its assessment of "text, structure, history, and so forth." Id. at 2416. In that hierarchy, text is paramount. Riccardi , 989 F.3d at 483 ; see City of Arlington , 569 U.S. at 301, 133 S.Ct. 1863 ("[T]he question in every case is, simply, whether the statutory text forecloses the agency's assertion of authority."). From these principles, and our caselaw applying them in the Guidelines context, we can identify at least three necessary properties of reasonable agency interpretations. The Commission's 75:1 rule lacks all three.

Property One : To be a "reasonable interpretation" of a Guideline, a rule contained in the commentary must first be an "interpretation" of that Guideline, not a "substantive policy choice." Riccardi , 989 F.3d at 487 ; accord United States v. Havis , 927 F.3d 382, 386 (6th Cir. 2019) (en banc). The distinction matters. "Unlike the Guidelines themselves, ... commentary to the Guidelines never passes through the gauntlets of congressional review or notice and comment." Havis , 927 F.3d at 386. "That is ... not a problem," so long as the commentary "serves only to interpret the Guidelines' text, not to replace or modify it." Id. (citations omitted); see also Kisor , 139 S. Ct. at 2420 (plurality) (rejecting argument that Auer allows agencies to evade notice-and-comment restrictions because agency interpretations are not legislative rules).

Is "1/75th of a video" a mere "interpretation" of the word "image?" No. The government and the majority opinion scarcely attempt to maintain that it is. The government acknowledges that, as a matter of ordinary meaning, a "video" is "[a] sequence of images processed electronically into an analog or digital format and displayed on a screen with sufficient rapidity as to create the illusion of motion and continuity." Appellee Br. at 26 (quoting American Heritage Dictionary 1930 (5th ed. 2016)). Against that backdrop, the Commission rejected the argument that one video equaled one image, explaining that "because each video contained multiple images it should be counted as more than one image." U.S. Sent'g Comm'n, The History of the Child Pornography Guidelines 43 (2009). The Commission also "knew that, according to the Motion Picture Association, a video contains 24 frames per second, and with each frame counted as a single image, a one minute video would contain 1,440 images." Appellee Br. at 26 (citing The History of the Child Pornography Guidelines 43 n.201). It thus considered that the number of images in a video might simply equate to the number of frames. Id. So far so good. But the Commission thought the consequences of that rule would be tough: Many, perhaps the vast majority, of child-pornography defendants would receive a 5-level enhancement. Id. So the Commission jettisoned the only definitions of "image" it had considered in favor of what the government describes as "a parsimonious rule," a "happy medium," a "conservative" rule. Id. at 26–28. Do those explanations suggest an act of "interpretation?" An attempt to "derive a proposition from an existing document whose meaning compels or logically justifies the proposition?" Cath. Health Initiatives v. Sebelius , 617 F.3d 490, 494 (D.C. Cir. 2010) (quotation marks and citation omitted). To ask is to answer.

If one needed more evidence, look no further than the commentary itself. There, the Commission all but admits that it's making a policy choice. Whereas the first part of Application Note 6 says that " ‘[i]mages’ means any visual depiction," U.S.S.G. § 2G2.2(b)(7) n.6(A) (emphasis added), the second says that "[e]ach video, video-clip, movie, or similar visual depiction shall be considered to have 75 images," id. § 2G2.2(b)(7) n.6(B)(ii) (emphasis added). "Shall be considered" is the language of a policy choice, not of interpretation. As the Second Circuit has explained, statutes use the phrase "shall be considered" to discard a term's "ordinary ‘plain English’ meaning" in favor of a "legal fiction" that "achieve[s] certain social policy goals." Sarmiento v. United States , 678 F.3d 147, 152 (2d Cir. 2012). The Second Circuit was in good company. The Supreme Court, in a unanimous opinion, has made this point too when discussing the synonymous phrase, "shall be deemed." See Sturgeon v. Frost , ––– U.S. ––––, 139 S. Ct. 1066, 1076, 203 L.Ed.2d 453 (2019). "Legislators (and other drafters)" use that phrase to create "a legal fiction," so that "abracadabra-style" a thing is legally treated as "what it is not." Id. at 1081 (cleaned up). So too here. Every video shall be considered to contain 75 images and so that's the number it has. Voilà! The Commission has conjured a construction, rather than construed a text.

Alas, because it did not approve of the consequences, the Commission forthrightly rejected any interpretation of images in favor of a "happy medium" that was more "[ ]fair." Appellee Br. at 25–26. Maybe the 75:1 rule is fair; maybe it's not. But the government's own explanation shows that it isn't an interpretation of the term "image." It is a "substantive policy choice" that "belongs in the guidelines, not in the commentary." Riccardi , 989 F.3d at 487 ; accord Havis , 927 F.3d at 386.

The majority opinion does not seriously try to defend the 75:1 rule as an "interpretation" either. Indeed, the opinion contains a single line about text, buried in a paragraph about purpose and results. Here it is: "The 75:1 Rule takes into account the text, acknowledging that a video is comprised of multiple images, as both terms are commonly understood." Maj. Op. at –––– – ––––. So, as the majority opinion sees it, the only textual limit on the Sentencing Commission's interpretative discretion is that it must ensure that a video counts as more than one image.

Because its reading of the text is so boundless, the majority opinion needs "structure, history, and so forth" to do work. See Kisor , 139 S. Ct. at 2416. The majority opinion says, "the Commission acknowledged the disproportionate results that would occur if it adopted either Phillips's one-video-one-image calculation or a one-frame-one-image calculation." Maj. Op. at ––––. But "disproportionate results," like "happy medium," is the language of a policy judgment, not an interpretation. See Chapman v. United States , 500 U.S. 453, 466–67, 111 S.Ct. 1919, 114 L.Ed.2d 524 (1991) (explaining that differentiating sentences is a question of "public policy enacted into statutes"). The place to address disproportionate results is the Guidelines, not the commentary. Riccardi , 989 F.3d at 487. The Commission's attention to the "scale of the differentiation" in the image table does not transform its 75:1 rule into an interpretation either. Contra Maj. Op. at –––– (emphasis omitted). The table's ranges no more "compel[ ] or logically justif[y]" the choice of 75 than, say, 25 or 50. See Cath. Health , 617 F.3d at 494 (citation omitted).

Property Two : The agency's interpretation must "line[ ] up with one of" the "reasonable meanings" ascertained in determining whether the Guideline is genuinely ambiguous. Kisor , 139 S. Ct. at 2419 (plurality). Riccardi demonstrates this point. After plumbing dictionaries for the meaning of the word "loss," we explained that even if there were multiple plausible options, the Commission's wasn't one of them. A meaning that "you will not find in any dictionary" is outside the "zone of ambiguity." 989 F.3d at 479, 485.

Is "1/75th of a video" one of the reasonable meanings of "image" derived from the statutory interpretation of § 2G2.2(b)(7) ? No. In its discussion of ambiguity, the majority opinion identifies three possible meanings of "image": (1) a video, (2) a frame, and (3) the ill-defined meaning it intends to convey with its Bergman hypothetical. Maj. Op. at –––– – ––––. Notably absent from that list is the 75:1 rule chosen by the Commission. Faced with a situation in which "[t]here can be no thought of deference," Kisor , 139 S. Ct. at 2419, the majority opinion nevertheless defers.

Property Three : "[A] specific numeric amount ... generally will not qualify as a mere ‘interpretation’ of general nonnumeric language." Riccardi , 989 F.3d at 487. Given its clear, binding application in this case, this rule hardly needs further justification. But I note that this court joins others in recognizing this fundamental principle. E.g. , Cath. Health , 617 F.3d at 495 ; Hoctor v. USDA , 82 F.3d 165, 169–71 (7th Cir. 1996) ; see also United States v. Bert , 292 F.3d 649, 652 (9th Cir. 2002) (noting that arriving at an "arbitrary quantitative determination" like the minimum amount of cocaine base that is a "detectable amount" is a "quintessentially legislative function"); Mo. Pub. Serv. Comm'n v. FERC , 215 F.3d 1, 4 (D.C. Cir. 2000) (explaining that numerical limits "cannot readily be derived by judicial reasoning" because "it is impossible to give a reasoned distinction between numbers just a hair on the OK side of the line and ones just a hair on the not-OK side").

Is "1/75th of a video" consistent with this rule? Again, no. And again, no one even pretends that it is. "Images" is "general nonnumeric language," just like "cost" was in Riccardi ; and just as the amount of "loss" in Riccardi would "turn on such fact-dependent things as the value of the gift card or the costs of replacing it," the number of "images" in a video will turn on "such fact-dependent things as" the length of the video and the number of frames displayed per second. 989 F.3d at 486. Nevertheless, the majority accedes to the government's request that we defer to a "fictional" "bright-line" rule that cannot "be derived ... by a process reasonably described as interpretation." Id. at 487 (quoting Hoctor , 82 F.3d at 170 ).

The majority opinion distinguishes Riccardi on the basis that the "loss" in Riccardi could be determined purely as a factual matter, whereas here the number of images in a video cannot be ascertained "without making some sort of interpretation." Maj. Op. at ––––. But an interpretation was required in Riccardi , too. We expressly noted that "loss" might mean "the value of the gift card or the costs of replacing it"—two different interpretations—because loss "might include the costs associated with obtaining a replacement gift card, including the time and expense from a second trip to the store." Riccardi , 989 F.3d at 486 (emphasis added). Either path would require a fact-dependent inquiry. Id. So too here. A court will first have to decide whether image means frame, a whole video, or something else. But after that, the number of images will depend on facts. This is necessarily true for nonnumeric language. Id. at 487.

The majority also suggests that Riccardi is different because there the Commission's "numerical formula" was "unreasonable." Maj. Op. at ––––. That distinction not only has no support in Riccardi 's text, but it also contradicts its core reasoning. It was the very act of choosing a number that revealed the substantive, rather than interpretive, character of the Commission's commentary. Riccardi , 989 F.3d at 487. Indeed, the suggestion that the Commission's "numerical formula" should be judged by its reasonableness effectively concedes that the formula is a policy decision, not an interpretation. See Cath. Health , 617 F.3d at 494–95 (explaining that assigning a particular numeric application to a term like "reasonable cost" is "not ordinarily [a process] of interpretation").

The upshot is that the Commission's 75:1 rule is not, by any measure, an interpretation of the Guideline. It is a substantive policy choice to which this court does not owe—indeed cannot owe—any deference. Riccardi , 989 F.3d at 487 ; Havis , 927 F.3d at 386. That inevitable conclusion suffices to reject the majority opinion and the Commission's rule. But it is not enough to resolve this case.

II.

To resolve this case, we still need to decide whether the district court was right to apply a five-level enhancement to Phillips's sentence under U.S.S.G. § 2G2.2(b)(7). To do so, we need to know what the right interpretation of "image" is. We know that it does not, and cannot, mean 1/75th of a video. But what does "image" mean?

Between the briefing and the majority opinion, three options remain: "Image" might mean (1) a video, (2) a frame, or (3) something like "imagery," "impression," or "scene." If option one, Phillips's preferred approach, is right, then his Guidelines range would be reduced. If option two—the government's fallback position—is correct, then Phillips's Guidelines range remains the same. If option three—the majority's conjecture based on a hypothetical viewing of a Scandinavian art-house film—prevails, then I cannot work out how Phillips's Guidelines range would be affected. The majority, tellingly, makes no attempt to explain either.

Happily, the answer is clear. An "image" means exactly what one would think: a "still representation" of something. And, in the context of a video, that means a "frame."

A.

To determine the plain meaning of a Guideline, we start with the text. United States v. Hill , 963 F.3d 528, 532 (6th Cir. 2020) ; see also Babb v. Wilkie , ––– U.S. ––––, 140 S. Ct. 1168, 1172, 206 L.Ed.2d 432 (2020). Usually, that means starting with a dictionary. Riccardi , 989 F.3d at 486 ; United States v. Zabawa , 719 F.3d 555, 559 (6th Cir. 2013). The Oxford English Dictionary provides several definitions of "image." First, it defines image, in a technical sense, as "[a] physical or digital representation of something, originally captured using a camera from visible light, and typically reproduced on paper, displayed on a screen, or stored as a computer file." Image , Oxford English Dictionary (3d ed. 2009), https://www.oed.com (last visited Aug. 23, 2022). Colloquially, it defines "image" as "any picture or graphic" displayed "in printed form." Id. Either way, "image" means a still representation. One can hardly reproduce a moving thing "on paper" or "in printed form." Pictures and graphics are both still. See Graphic , Oxford English Dictionary, supra (defining "graphic" as something drawn, painted, engraved, or etched). And other definitions agree that an image is a singular, still representation of something. See Image , Oxford English Dictionary, supra ("[A] likeness, portrait, picture, carving, or the like."); Image , Merriam Webster's Collegiate Dictionary 619 (11th ed. 2003) ("[A] likeness of an object" or "a picture."); Image , The American Heritage Dictionary of the English Language 877 (5th ed. 2018) ("A representation of the form of a person or object, such as a painting or photograph."); Image , Webster's New World College Dictionary 726 (5th ed. 2020) ("[A] representation or likeness of a person or thing, as in a drawing, painting, photograph, or sculpture."); Image , Shorter Oxford English Dictionary 1327 (6th ed. 2007) ("A representation of the external form of a person or thing in sculpture, painting, etc."). This definition isn't really contested, even by Phillips.

So, turning to the facts of this case, we could rephrase the question as, "how many still representations did Phillips possess in his 91 videos?" The answer comes readily to anyone who has ever owned a remote with a pause button: just count the number of frames. In fact, "frame" is just what we call an image when that image is used in a video. See Frame , Oxford English Dictionary, supra ("[O]ne of the individual images on a strip of film" or "a single complete image in a series forming a television picture, film, or video sequence."); Merriam Webster, supra , at 497 ("[O]ne picture in the series on a length of film" or "a complete image for display."); Frame , American Heritage, supra , at 695 ("One of the set of still images that constitute a film or video."); Frame , Webster's New World, supra , at 574 ("[T]he rectangular image on a film screen."); Frame , Oxford Shorter, supra , at 1032 ("A single complete image or picture built up from a series of lines."). Could anyone really argue that if Phillips paused one of his videos and printed a frame, that wouldn't count as an image? Or that if Phillips saved each frame individually, each wouldn't count as one image? I don't think so. And the Commission doesn't either. It defined "image[ ]" to include anything which is "capable of conversion into a visual image." See U.S.S.G. § 2G2.2(b)(7) n.6(A) (incorporating the definition contained in 18 U.S.C. § 2256(5) ).

"Courts presume that an undefined word comes with its ordinary meaning," Riccardi , 989 F.3d at 488, and that ordinary meaning isn't even contested here. "Image," in the context of a video, means "frame." Neither Phillips's alternative definition nor the majority's can unseat this clear frontrunner.

B.

Phillips's preferred definition—"image" equals "video"—can be quickly dismissed, and the majority opinion rightly does so. It is contradicted by every dictionary definition and by common sense. A video is "a recording of visual images and sound." Merriam Webster, supra , at 1394 (emphasis added); see also Video , Oxford English Dictionary, supra (Videos "relat[e] to, or [are] concerned with the images displayed on a television or other electronic device, or the electrical signal, channel, etc., conveying such images ." (emphasis added)); Video , American Heritage, supra at 1930 ("A sequence of images processed electronically into an analog or digital format and displayed on a screen with sufficient rapidity as to create the illusion of motion and continuity."); cf. MCI Telecomm. Corp. v. AT&T Co. , 512 U.S. 218, 227, 114 S.Ct. 2223, 129 L.Ed.2d 182 (1994) (rejecting an alternative definition of "modify" that "not only supplements the meaning contained in all other dictionaries, but contradicts one of the meanings contained in virtually all other dictionaries").

With all the evidence pointing the other way, Phillips argues that "images" is a term of art in the statutory context that really means "visual depiction." But, as the majority rightly points out, those terms have different meanings, and Congress explicitly replaced the latter with the term "images." Maj. Op. at –––– – ––––. With Phillips's definition eliminated, the only other interpretation offered by the parties is "frame."

C.

The majority opinion recognizes the support for "frames" in the text. Maj. Op. at ––––. But it finds ambiguity in an unsupported, paragraph-long hypothetical about two friends and an Ingmar Bergman movie, The Seventh Seal . The problems in this paragraph abound, most noticeably that the hypothetical contains no ascertainable theory of what constitutes an "image."

In the hypothetical, one friend asks if The Seventh Seal "contained any striking ‘images,’ " to which the other friend responds, "Yes, I found the iconic image of Death playing chess with the crusader on a deserted beach very memorable." Id. at –––– – ––––. The hypothetical ends with the friends struggling to pinpoint in which frame the image appears because "chess-playing Death does not correspond to any particular frame, nor to any singular still moment at all." Id. at ––––. Figuring out what kind of "image" the majority opinion means is a puzzle. It's clear that the image is somehow moving because it's more than a "still moment." But it's not the whole movie; it's just the part that makes up "chess-playing Death." Does the majority opinion mean imagery? Impression? Scene? It never tells us, probably because none of these things could be an "image" as it is used in the image table in § 2G2.2(b)(7).

"Imagery"—"the representation of ideas with images; visual metaphor, symbolism"—seems to approach what the majority is getting at with its "chess-playing Death" hypothetical. Imagery , Oxford English Dictionary, supra . I have not seen the film, but I imagine that Bergman meant the chess match to represent something like, "Man's futile quest to outwit the fact of his own mortality." However well this sort of "imagery" works on the Big Screen, it makes no sense in the child-pornography Sentencing Guideline. First, how many videos—or still photographs—that capture acts of child pornography use imagery? My guess is vanishingly few. At least 10 "visual metaphors" would be required for any enhancement at all, and more than 600 to receive the maximum punishment. So the vast majority of offenders, no matter how many photographs or 3-hour videos they possessed, would receive no enhancement, rendering useless the "image table's precision." Majority Op. at ––––. More fundamentally, the majority offers absolutely no support from the "text, structure, history, and so forth," Kisor , 139 S. Ct. at 2416, for its startling suggestion that Congress's sole aim in the image table was to punish child pornography more harshly when the producer repeatedly used "visual metaphors" or "symbolism," a choice that seems to bear little relationship to the very real harms done to child-pornography victims.

What about "impression"? "Chess-playing Death" might leave "[a]n effect, especially a strong effect, produced on the intellect, conscience, or feelings." Impression , Oxford English Dictionary, supra . The friend in the hypothetical found it "very memorable," after all. But "impression" plainly doesn't mean "image" either. An image might leave an impression, or it might not. And surely Congress intended to incrementally punish every image of child pornography, not just those that had a "strong effect." The remainder of § 2G2.2 includes enhancements based on the offender's conduct and the content of the materials possessed. E.g. , § 2G2.2(b)(2) (material involving prepubescent minor); § 2G2.2(b)(4) (sadistic or masochistic material). Are we to believe Congress suddenly decided in the image table to instead focus on whether the material left an impression on the viewer?

How about "scene"? "Chess-playing Death" might describe "[a] sequence of dramatic action." Scene , Oxford English Dictionary, supra . But that has hardly anything to do with "image"; pictures and graphics, which are obviously images, don't have a sequence of dramatic action, unless strung together. And do we think Congress meant for district courts to comb through child-pornography videos, looking for sequences of dramatic action, or, for that matter, tallying their "impressions" or the uses of "imagery"?

As further evidence that "imagery," "impression," and "scene" make no sense in the context of the image table, consider the following: It's clear how an offender can possess or distribute a picture or graphic, or frames in a video. But how can an offender "possess" an "impression" or "imagery"? And if "image" means "scene," are photographs no longer included in the image table? A photograph is clearly not a "scene," however it's defined.

One suspects, of course, that the majority doesn't actually think that "image," as used in the Guideline, could reasonably mean "imagery," "impression," or "scene"—or it would say so outright. Instead, the opinion just waves its hand in the direction of uncertainty so it can conclude that "the number of frames in a video ... is not necessarily the only way to understand the image table even as a textual matter." Maj. Op. at ––––. But the majority opinion has made no effort to show how its vague allusion to "images" in a Scandinavian art-house film could work in the image table. Nor has it tried to show how the Bergman version of "image" stacks up against "frame." We are to "exhaust all the ‘traditional tools’ of construction" before we cry "ambiguous." Kisor , 139 S. Ct. at 2415. But the majority opinion offers no competing dictionary, no canon of construction, and not a single comparison between definitions, to see which is better supported by the text, structure, history, or purpose. This cannot be the "taxing inquiry" Kisor demands of us. 139 S. Ct. at 2415 (quoting Pauley v. BethEnergy Mines, Inc. , 501 U.S. 680, 707, 111 S.Ct. 2524, 115 L.Ed.2d 604 (1991) (Scalia, J., dissenting)).

D.

The majority opinion's real problem with reading "image" to mean "frame" isn't deciphering the text; instead, the majority worries that if it gives the text its natural reading, it will lead to "decidedly unreasonable results." Maj. Op. at ––––. The majority does not explain its "decidedly unreasonable results" rule, and I am unfamiliar with it as a tool of interpretation. The closest I can come is the "absurdity" doctrine, which posits that courts may depart from the text when it would produce "an outcome so contrary to perceived social values that Congress could not have ‘intended’ it." John F. Manning, The Absurdity Doctrine , 116 Harv. L. Rev. 2387, 2390 (2003). But the absurdity canon, if it ever applies, is for "rare and exceptional circumstances" when the absurdity is "so gross as to shock the general moral or common sense." Crooks v. Harrelson , 282 U.S. 55, 60, 51 S.Ct. 49, 75 L.Ed. 156 (1930). Only when "it is quite impossible that Congress could have intended the result," and when that conclusion must be "obvious to most anyone," may we apply the canon. Pub. Citizen v. DOJ , 491 U.S. 440, 471, 109 S.Ct. 2558, 105 L.Ed.2d 377 (1989) (Kennedy, J., concurring in the judgment). An inch short of that and we are simply discarding the text for our own policy preferences. In re Hokulani Square, Inc. , 776 F.3d 1083, 1088 (9th Cir. 2015) ; see also Griffin v. Oceanic Contractors, Inc. , 458 U.S. 564, 576, 102 S.Ct. 3245, 73 L.Ed.2d 973 (1982).

For three reasons, the majority opinion's analysis falls well shy of the absurdity bar.

1.

First, the majority opinion bemoans what it deems the harsh results that would befall the (hopefully rare) defendant who possesses slow-motion child pornography or who films Ken Burns-esque videos of his victims. Maj. Op. at –––– – ––––. But even if the results are harsh, the caselaw is clear: Harsh results are not "absurd" results that we may set aside. Lamie v. U.S. Tr. , 540 U.S. 526, 538, 124 S.Ct. 1023, 157 L.Ed.2d 1024 (2004) ("Our unwillingness to soften the import of Congress' chosen words even if we believe the words lead to a harsh outcome is longstanding."); Dodd v. United States , 545 U.S. 353, 359, 125 S.Ct. 2478, 162 L.Ed.2d 343 (2005) ("Although we recognize the potential for harsh results in some cases, we are not free to rewrite the statute that Congress has enacted .... [T]he text here, though strict, is not absurd."); St. Louis, Iron Mountain & S. Ry. Co. v. Taylor , 210 U.S. 281, 295, 28 S.Ct. 616, 52 L.Ed. 1061 (1908) ("It is urged that this is a harsh construction. To this we reply that, if it be the true construction, its harshness is no concern of the courts."); Chapman , 500 U.S. at 458, 463–64, 111 S.Ct. 1919 (holding that an LSD sentencing statute did "not produce a result ‘so "absurd or glaringly unjust" ’ as to raise a ‘reasonable doubt’ about Congress' intent" even though it could produce results both "anomalous" and "irrelevant to culpability," "viz: a major wholesaler caught with 19,999 doses of pure LSD would not be subject to the 5-year mandatory minimum sentence, while a minor pusher with 200 doses on blotter paper, or even one dose on a sugar cube, would be subject to the mandatory minimum sentence"); United States v. Rodgers , 466 U.S. 475, 482, 484, 104 S.Ct. 1942, 80 L.Ed.2d 492 (1984) (holding that a criminal statute was not "absurd" even though, "giv[ing] the statute a ‘literal interpretation’ " resulted in harsher punishment for one who "informally volunteer[ed] an untrue statement to an F.B.I. agent than" for one who "relate[d] the same story under oath before a court of law"); In re Blixseth , 684 F.3d 865, 872 (9th Cir. 2012) ("The result may be harsh but is not absurd."); Hokulani Square , 776 F.3d at 1088 ("The distinction drawn by [the statute] may be harsh and misguided, but it is not absurd."); Dekoladenu v. Gonzales , 459 F.3d 500, 506 (4th Cir. 2006) ("Although this result may be harsh, it is hardly ‘nonsensical’ or ‘absurd.’ ").

The justification for this rule is simple. The absurdity canon "isn't a license for us to disregard statutory text where it conflicts with our policy preferences," Hokulani Square , 776 F.3d at 1088, but whether a punishment is "harsh" is necessarily a policy judgment. As Judge Easterbrook once explained: "Laws are not ‘harsh’ or ‘pointless’ in any value-free framework; they seem harsh or pointless by reference to a given judge's beliefs about how things ought to work, which is why a claim of power to revise ‘harsh’ or ‘pointless’ laws elevates the judicial over the legislative branch and must be resisted." United States v. Logan , 453 F.3d 804, 806 (7th Cir. 2006), aff'd , 552 U.S. 23, 128 S.Ct. 475, 169 L.Ed.2d 432 (2007) ; see also Crooks , 282 U.S. at 60, 51 S.Ct. 49.

2.

Even if we were to play the harsh-equals-absurd game, it's not at all clear that the majority opinion would win. The majority opinion points out that if "image" means "frame," that "would nearly automatically vault [most] offender[s] to the top of the range," undermining the purpose of the tiers to differentiate. Maj. Op. at ––––. But in child-pornography cases, a lack of differentiation isn't even unusual, much less absurd. Of all sentences under § 2G2.2 in 2020, 93.6% involved an image of a prepubescent minor (qualifying for a two-level increase); 94.2% involved a computer (qualifying for a two-level increase); and 57% involved an image depicting sadistic or masochistic conduct or other forms of violence (qualifying for a four-level enhancement). See U.S. Sent'g Comm'n, Use of Guidelines and Specific Offense Characteristics for Fiscal Year 2020 , at 42. In fact, under the current rule the vast majority of offenders—73%—possessed 600 or more images and qualified for a five-level enhancement. Id. And 93.6% qualified for some enhancement under the image table. Id. Thus, construing image to mean "frame" leads to results that are, in reality, well within the Guidelines norms.

Moreover, our cases have repeatedly "rejected the idea" that the child-pornography enhancements "should be ignored because [they] so frequently appl[y]." United States v Holland , 799 F. App'x 380, 385 (6th Cir. 2020) ; see United States v. Walters , 775 F.3d 778, 786 (6th Cir. 2015) ; United States v. Cunningham , 669 F.3d 723, 732–33 (6th Cir. 2012). So the outcome of applying the plain text here is a far cry from the kind of results we, or the Supreme Court, have considered absurd. See, e.g. , Green v. Bock Laundry Mach. Co. , 490 U.S. 504, 528, 109 S.Ct. 1981, 104 L.Ed.2d 557 (1989) (Scalia, J., concurring in the judgment) (providing "the benefit of prejudice-weighing to civil defendants and not civil plaintiffs"); United States v. Kirby , 74 U.S. 482, 487, 7 Wall. 482, 19 L.Ed. 278 (1868) (charging a prisoner who breaks out of prison "when the prison is on fire"); Pub. Citizen , 491 U.S. at 471, 109 S.Ct. 2558 (1989) (Kennedy, J., concurring in the judgment) (applying a "medieval law against drawing blood in the streets" to "a physician who came to the aid of a man who had fallen down in a fit"); United States v. Underhill , 813 F.2d 105, 112 (6th Cir. 1987) (using a statute "protect[ing] the victims of unlawful interceptions ... to shield the very people who committed the unlawful interceptions from the consequences of their wrongdoing").

Even if the textual reading did further dilute differentiation between offenders, the majority opinion overstates the problem by suggesting that it would "obviat[e] the purpose of prescribing different levels." Maj. Op. at ––––. It is far from absurd to think that Congress would want to carefully distinguish offenders possessing still photographs but give the maximum enhancement to most offenders possessing videos. Indeed, the Department of Justice has recognized "the increased harm caused by moving videos." Letter from the Office of the Attorney General to Members of the U.S. Sentencing Commission (Mar. 1, 2004); cf. Holland , 799 F. App'x at 385 ("[T]he prevalence of computers in the commission of [child-pornography] offenses, far from undermining the enhancement, is its very raison d'etre .").

Nonetheless, the majority opinion concludes that Congress could not have intended "to penalize videos to that degree." Maj. Op. at ––––. If the majority means this as a normative statement ("no fairminded Congress would have penalized videos to that degree"), that judgment is not ours to make. If the majority intended a predictive statement ("Congress could not have meant frames per second because it's so out of step with how Congress has behaved in the past"), the majority is reasoning from false premises. For better or worse, a severe punishment for most offenders is likely exactly what Congress intended. See United States v. McNerney , 636 F.3d 772, 776–78 (6th Cir. 2011) (cataloguing Congress's repeated amendments to the child-pornography guidelines to increase punishment); United States v. Bistline , 665 F.3d 758, 764 (6th Cir. 2012) (pointing out that "Congress's long and repeated involvement in raising the offense levels for § 2G2.2 makes clear that the grounds of its action were not only empirical, but retributive—that they included not only deterrence, but punishment"); see also U.S. Sent'g Comm'n, Federal Sentencing of Child Pornography: Non-Production Offenses , 1–3 (2021) (explaining that the Commission in 2012 recommended that Congress amend § 2G2.2 to address the lack of differentiation, but noting that "[t]o date, Congress has not implemented the Commission's ... recommendations"); U.S. Sent'g Comm'n, The History of the Child Pornography Guidelines 54 (2009) (noting Congress's "continued interest in deterring and punishing child pornography offenses," including through laws that "increased criminal penalties").

3.

Finally, put aside the fact that harshness and absurdity are not the same, and that construing "image" to mean "frame" does not actually lead to anything remotely close to results that this court or the Supreme Court has recognized as absurd. Instead ask whether the majority opinion's solution (deferring to the 75:1 rule) would even solve the differentiation problems that it believes justify eschewing the plain meaning of the text? The answer is no.

Consider, for example, that under the 75:1 approach, an offender with a three-second video is treated the same as an offender with a three-minute video—and, possibly, the same as a three-hour video. Each of these videos "shall be considered" to have 75 images. It's not ridiculous to think that some offenders will have very short videos; indeed, Phillips himself possessed a three-second video with a framerate of 15. R. 126, PageID 455. Under the plain text of the Guideline, that video would count as 45 images. Under the majority's rule, it counts as 75. So for a defendant with short videos, the 75:1 rule will be harsher than a frames-based rule. The only reason a three-hour video might be treated differently is the escape-valve provision, but when it applies, adherence to "the goal of differentiating penalties," Maj. Op. at ––––, would stem not from the Guideline itself, but from the sentencing court's discretion.

The problem is exacerbated by the loophole created by the 75:1 rule. A defendant who compiles thousands of pictures into a single video as a slideshow would receive the minimum enhancement, but a defendant who possesses those same pictures in printed form or as individual files would receive the maximum. Surely tech savvy child-pornography defendants with the capability to aggregate images in this way are at least as common as the documentarian or avant-garde defendants the majority opinion seeks to protect. See Maj. Op. at ––––, –––– – ––––. Sentencing rules usually avoid differentiating based on format because of "the ease of conversion between formats and relative unimportance of the format to the user's ability to view the image" in child-pornography cases, United States v. Ladeau , No. 09-40021-FDS, 2010 WL 1427523, at *6 (D. Mass. Apr. 7, 2010), yet the 75:1 rule accepts those variations without a second thought. A frame is an image. It can be viewed like an image. It can be saved like an image. It can be printed like an image. It, in fact, fits Congress's own definition of "image," which includes data "capable of conversion into a visual image." 18 U.S.C. § 2256(5). And yet the 75:1 rule differentiates not based on the content or quality of images, but on the file type the possessor happens to choose.

The majority opinion's specific hypotheticals do not move the needle. The opinion compares an offender who has "100 photographs, each depicting a different victim," to an offender who "has a single 30-second video depicting a single victim," and laments that equating "image" with "frame" would cause the former to get a two-level increase and the latter a five-level increase. But, under the 75:1 rule, an offender with 600 copies of the same photograph of one victim would get a five-level increase, McNerney , 636 F.3d at 780, while an offender with a five-minute video of hundreds of different minors and situations would get a two-level increase. The majority opinion also asks us to consider that, by applying the plain text, possessing "a video gently panning over a photograph" of child pornography would be treated more harshly than simply possessing the same photograph. Maj. Op. at ––––. But what about a video quickly panning over thousands of photographs? The 75:1 rule treats possessing that film more mercifully than possessing the exact same photographs, even though the offender can simply pause the video at each photograph. The majority opinion is also concerned about cases in which an "offender films a five-second scene with a regular camera, while another offender films that same scene with a slow-motion camera," and the latter is punished much more severely. Maj. Op. at –––– – ––––. Such an obscure hypothetical hardly warrants rejecting the plain text of the Guidelines. But in any event, the 75:1 rule shares the same faults. What about a case in which one offender films a five-second scene with a regular camera, and the other offender films for hours, compiling it all into a sped-up five-minute video? Under the 75:1 rule, both offenders are treated the same. It's no more outlandish to think that child-pornography distributers will speed up videos than slow them down.

So are the results of the 75:1 rule also "totally unmoored from the goal of differentiating penalties based on the amount of child pornography?" Id. at ––––. It would seem so. But the majority opinion ignores that absurdity in its haste to defer to the Sentencing Commission.

* * *

I recognize that people may disagree over whether the frame-based rule, the 75:1 rule, or some other rule, would better serve the goal of differentiating offenders based on their culpability. We might debate how commonly child-pornography defendants possess short videos, whether offenders who film their victims with a slow-motion camera should be punished more severely, or even whether possessing a video is categorically worse than possessing still photos, and by how much. But what seems beyond peradventure is that there is room for disagreement. And once that's true, the absurdity canon has no place in a court's analysis. Pub. Citizen , 491 U.S. at 471, 109 S.Ct. 2558 (Kennedy, J., concurring in the judgment). When a court discards the plain meaning of a text on the basis of anything less than an "impossible" result, id. , as the majority has done here, it makes a policy choice, and "elevates the judicial over the legislative branch," Logan , 453 F.3d at 806.

E.

To recap: Once the 75:1 rule is rejected, three candidates for the definition of "image" remain—video, frame, or imagery/impression/scene. Of these three, only "frame" has any support in the text of the Guideline and neither the government nor the majority opinion deploys any tool of statutory interpretation to reject "frame." The majority opinion makes no effort to show that its Ingmar-Bergman-inspired definition could reasonably apply. And any attempt to create ambiguity based on absurdity is self-defeating, not to mention a flagrant violation of precedent. For all that handwringing, we're right back where we started: "Images" means exactly what you'll find in every dictionary—a "still representation"; and vis-à-vis a video, an "image" is a "frame." Because "there is only one reasonable construction" of the Guideline, the "court has no business deferring to any other reading, no matter how much the [Commission] insists it would make more sense." Kisor , 139 S. Ct. at 2415. The majority opinion's (hopefully anomalous) rush to defer returns us to the world of old Auer . I cannot join it in doing so.

* * *

The word "image" as used in U.S.S.G. § 2G2.2(b)(7) means frame. By that measure, Trinity Phillips possessed well over 600 images and so the maximum enhancement applies. I concur in the judgment only.