United States v. Flores-Gonzalez

Opinion

United States Court of Appeals For the First Circuit

No. 19-2204 UNITED STATES OF AMERICA, Appellee,

v.

EMILIANO EMMANUEL FLORES-GONZÁLEZ, Defendant, Appellant.

APPEAL FROM THE UNITED STATES DISTRICT COURT FOR THE DISTRICT OF PUERTO RICO

[Hon. Francisco A. Besosa, U.S. District Judge]

Before

Barron, Chief Judge,

Lynch, Thompson, Kayatta, Gelpí, and Montecalvo, Circuit Judges.

Kevin E. Lerman, Research and Writing Attorney, with whom Eric Alexander Vos, Federal Public Defender, Franco L. Pérez- Redondo, Assistant Federal Public Defender, Supervisor, Appeals Section, and Alejandra Bird-López, Research and Writing Attorney, were on brief, for appellant.

Emma A. Andersson, Devi M. Rao, Elizabeth A. Bixby, and Fermin Arraiza on brief for Roderick & Solange MacArthur Justice Center, The American Civil Liberties Union Foundation, and The Puerto Rico Chapter of the American Liberties Union Foundation, amici curiae.

Adam Murphy, Janai S. Nelson, Samuel Spital, Ashok Chandran, Catherine Logue, and Christopher Kemmitt on brief for NAACP Legal Defense and Educational Fund, Inc., amicus curiae.

Judith H. Mizner, Assistant Federal Defender, on brief for Office of the Federal Defender for the Districts of Massachusetts, New Hampshire, and Rhode Island, amicus curiae.

Linda Backiel on brief for Puerto Rico Association of Criminal Defense Lawyers, amicus curiae.

Gregory B. Conner, Assistant United States Attorney, with whom W. Stephen Muldrow, United States Attorney, Mariana E. Bauzá- Almonte, Chief, Appellate Division, Kenneth A. Polite, Jr., Assistant Attorney General, Lisa H. Miller, Deputy Assistant Attorney General, John M. Pellettieri, Attorney, Appellate Section, and Jenny C. Ellickson, Attorney, Appellate Section, and were on brief, for appellee.

Opinion En Banc

November 7, 2023

The judgment entered in the district court is affirmed by an equally divided en banc court. See Savard v. Rhode Island, 338 F.3d 23, 25 (1st Cir. 2003) (en banc).

Opinions follow.

KAYATTA, Circuit Judge, with whom LYNCH and GELPÍ, Circuit Judges, join. On this appeal, Emiliano Emmanuel Flores- González ("Flores") raises two challenges to his sentence following his guilty plea to a charge of illegally possessing a machine gun in violation of 18 U.S.C. § 922(o) -- first, that he was erroneously classified as a "prohibited person," and second, that his sentence was both procedurally and substantively unreasonable. All members of the panel that first heard this appeal and all members of the en banc court agree that Flores's classification as a "prohibited person" under U.S.S.G. § 2K2.1(a)(4)(B) was not clear error and for that reason is affirmed, as more fully explained in the separate opinion that follows this opinion.

What divides our court is how to rule on Flores's challenge to the district court's decision to vary upward eighteen months from the upper end of the guidelines sentencing range. We explain in this opinion why three members of the court conclude that the upward variance was within the district court's discretion.

4

I.

We begin by explaining what the district court did at sentencing. After hearing from counsel for each party, and considering the pre-sentencing report of probation, the district court calculated a guidelines sentencing range of twenty-four to thirty months. All agree that this calculation was free from error.

The district court also considered the full array of sentencing factors set forth in 18 U.S.C. § 3553(a). In so doing, the district court began by referencing the government's assertion that Puerto Rico was a hotspot for violence and stating that "crime in Puerto Rico far exceeds the known limits on the mainland." Flores took no objection to this assertion. The district court then discussed at length its perception that, given the "pervasive" occurrence of gun crimes in Puerto Rico, the impact of possessing a machine gun in Puerto Rico was "more serious than that considered by the Sentencing Commission when it drafted the guidelines." The court also explained that deterring the "population at large" from engaging in such behavior was an important factor in sentencing.

The court then continued to discuss the specific characteristics of Flores and the characteristics of the offense. The court observed that, at the time of his arrest (at a McDonalds), Flores had the machine gun loaded with thirty-three rounds of ammunition, and he possessed an additional thirty rounds.

5

An empty shell casing was also found in the vehicle in which Flores had been riding at the time of his arrest. While mentioning these facts, the court did not claim that Flores's offense was more harmful than "others similar to his." Rather, the court's judgment was that gun crimes were more serious in Puerto Rico because of the scourge of violent crime being experienced in the Commonwealth. The court discussed the harm posed by machine guns, showing a video of a machine gun assault to illustrate the point. Citing a need for greater deterrence and punishment than was implicit in the guidelines range, the court varied upward by eighteen months to impose a sentence of forty-eight months. It is that variance that is at issue on this appeal.

II.

We certainly agree that a sentencing judge should focus carefully on the individual circumstances of the offender and the offense. The district court did exactly that, and said that it had done so. It is equally clear, too, that such a focus can properly encompass the location where the offense occurred, and that an offense can be seen as more serious (and necessitating greater deterrence) when committed in a community experiencing a greater-than-customary incidence of related crime. Our circuit has so held for well over a decade in as many as twenty-five cases.1

1 United States v. Politano, 522 F.3d 69, 74 (1st Cir. 2008);

United States v. Flores-Machicote, 706 F.3d 16, 22-23 (1st Cir.

6

It is also beyond debate that the need for general deterrence is a lawful consideration in sentencing. Section 3553(a) expressly commands courts to consider the need "to afford adequate deterrence to criminal conduct." 18 U.S.C. § 3553(a)(2)(B). And it is black letter law that the "criminal conduct" to be deterred by criminal sentences includes the conduct of persons other than the defendant, i.e., general deterrence. See United States v. Pagán-Walker, 877 F.3d 415, 417 (1st Cir. 2017) ("[T]he need for general deterrence is a permissible factor to consider [in sentencing].") The Supreme Court, too, has been

2013); United States v. Santiago-Rivera, 744 F.3d 229, 232-33 (1st Cir. 2014); United States v. Narváez-Soto, 773 F.3d 282, 286-87 (1st Cir. 2014); United States v. Rivera-González, 776 F.3d 45, 50-51(1st Cir. 2015); United States v. Zapata-Vázquez, 778 F.3d 21, 23-24 (1st Cir. 2015); United States v. Díaz-Arroyo, 797 F.3d 125, 129-30 (1st Cir. 2015); United States v. Pantojas-Cruz, 800 F.3d 54, 57, 59-60 (1st Cir. 2015); United States v. Paulino- Guzman, 807 F.3d 447, 450–51 (1st Cir. 2015); United States v. Bermúdez-Meléndez, 827 F.3d 160, 166 (1st Cir. 2016); United States v. de Jesús, 831 F.3d 39, 43 (1st Cir. 2016); United States v. Figueroa-Quiñones, 657 F. App'x 9, 13 (1st Cir. 2016) (unpublished); United States v. Santa-Otero, 843 F.3d 547, 551-52 (1st Cir. 2016); United States v. Vázquez, 854 F.3d 126, 129-30 (1st Cir. 2017); United States v. Fuentes-Echevarria, 856 F.3d 22, 26 (1st Cir. 2017); United States v. Lugo-Cartagena, 701 F. App'x 6, 11 n.5 (1st Cir. 2017) (unpublished); United States v. Garay- Sierra, 885 F.3d 7, 15-16 (1st Cir. 2018); United States v. Laureano-Pérez, 892 F.3d 50, 52-53 (1st Cir. 2018); United States v. Hernández-Ramos, 906 F.3d 213, 214-15 (1st Cir. 2018); United States v. Severino-Pacheco, 911 F.3d 14, 22 (1st Cir. 2018); United States v. Viloria-Sepulveda, 921 F.3d 5, 10 (1st Cir. 2019); United States v. Miranda-Díaz, 942 F.3d 33, 42-43 (1st Cir. 2019); United States v. García-Mojica, 955 F.3d 187, 193 (1st Cir. 2020); United States v. Díaz-Rivera, 957 F.3d 20, 29-30 (1st Cir. 2020); United States v. Gonzalez-Flores, 988 F.3d 100, 102-03 (1st Cir. 2021).

7

clear on the importance of general deterrence in sentencing. See Pell v. Procunier, 417 U.S. 817, 822 (1974) ("An important function of the corrections system is the deterrence of crime. The premise is that by confining criminal offenders in a facility where they are isolated from the rest of society, a condition that most people presumably find undesirable, they and others will be deterred from committing additional criminal offenses." (emphasis added)).

Our colleagues who write separately claim not to dispute the foregoing. In other words, they never actually say that a sentencing judge cannot consider the relative prevalence of gun crimes in the community in which the defendant decided to commit a serious gun crime. Instead, they rely on two recent panel opinions that create out of thin air two procedural limitations that -- as applied by our colleagues -- effectively eliminate any ability to use upwardly variant sentences in an effort to help a community experiencing a high level of gun crimes. See United States v. Rivera-Berrios, 968 F.3d 130 (1st Cir. 2020) and United States v. Carrasquillo-Sanchez, 9 F.4th 56 (1st Cir. 2021). We voted to proceed en banc in order to overrule those panel decisions to the extent they adopted such limitations. Our colleagues would instead affirm and apply those limitations in this case.

First, our colleagues would hold that a greater need to deter crime in a given community cannot serve by itself ("solely") to support any upward variance. There is absolutely no support

8

for this requirement in the text of section 3553(a) or in any Supreme Court opinion. And what exactly does it mean? The district court in this case considered a full range of sentencing factors, each of which presumably had some potential upward or downward effect. The court considered the locus of the crime and the resulting need for greater deterrence "solely" only in the sense that, but for that factor added to the rest, the court would have imposed a shorter sentence. This is exactly how sentencing works. To hold otherwise would simply be a back door way of saying that a court cannot upwardly vary based on a finding that the circumstances of the community in which the offense occurred render the offense more serious and the need for deterrence greater.

As for the second limitation, our colleagues say that the deterrence needs of a given community cannot support "too much" of an upward variance. We readily agree. But when one looks at the mandate our colleagues would issue on remand -- ordering no variance at all -- it becomes clear that "too much" means anything greater than zero. Their limitation would render it, effectively, procedural error for a sentencing court to impose an upward variance based on community characteristics. We think, instead, that "too much" in this context more properly means that a variance imposed because of community characteristics must still meet the requirements of substantive reasonableness, as we will explain. But we see no procedural error in the consideration of community

9

characteristics as they relate to section 3553(a) factors, including, as here, general deterrence.

This effort to enshrine judicially-created limitations that effectively overturn more than a decade of circuit precedent and preclude district court judges from providing added deterrence in aid of a community facing a relatively greater incidence of gun violence runs headlong into the Supreme Court's warning that "[t]he only limitations on a court's discretion to consider any relevant materials at an initial sentencing . . . are those set forth by Congress in a statute or by the Constitution." Concepcion v. United States, 142 S. Ct. 2389, 2400 (2023). There is no statutory or constitutional provenance for the limitations favored by our colleagues. Nor do they even claim to cite any.

Our decisions allowing sentencing courts to reflect a community's increased need to deter crime in an upwardly variant sentence makes especially good sense given the limitations of the guidelines. The Sentencing Commission recognizes that the need to deter constitutes an important consideration in calibrating the length of a sentence. See Rita v. United States, 551 U.S. 338, 347-49 (2007) (explaining that the guidelines are designed to carry out the sentencing objectives reflected in the section 3553(a) factors -- including, of course, deterrence). The Commission's calibration, however, is national, and thus unable to reflect local

10

variation. The guidelines also fail to change with a frequency that can capture some changes in our communities.2 Our colleagues who would vacate the sentence also say much about the need for a sentence to be "individualized." Indeed, their opinion repeats the various forms of the word at least twenty-five times, plainly implying that their approach to this appeal is "individualized" while the district court's was not.

They have it backwards. The district court simply considered and found significant one additional fact about the defendant and his offense that the guidelines did not take into account -- his commission of the offense in a community experiencing a high level of gun violence. As we observed previously, "community-specific characteristics" within that district "made [the defendant's] offense more serious and the need for deterrence greater than that reflected by the Guidelines." United States v. Politano, 522 F.3d 69, 73 (1st Cir. 2008). In that respect, it is the guidelines rather than the district court that paid less heed to the specifics of the offense.

At base, our colleagues' position rests on a policy judgment that because the guidelines do not account for local

2 Whether a district court can justify a variance such as in this case under Kimbrough v. U.S., 552 U.S. 85, 106-110 (2007), by expressing a policy disagreement with the Commission's one-sizefits -all approach, is not presented on this appeal. Nor was it addressed in Rivera-Berrios or Carrasquillo-Sanchez. See 9 F.4th at 61, n.2.

11

variations in the perceived need for deterrence, those variations cannot serve as a sufficient justification for treating the guidelines as advisory. And their repeated reference to general deterrence as a "questionable" rationale suggests that their policy judgment runs deep. Whether one agrees with this judgment is not the point. The point, instead, is that no court of appeals has the authority to impose -- as policy -- the sentencing limitations favored by our colleagues. Concepcion, 142 S. Ct. at 2400. Or, as we said previously, "Post-Booker, it is now apparent that the district court has the discretion to take into account all of the circumstances under which [the defendant] committed the offense, including the particular community in which the offense arose." Politano, 522 F.3d at 74. Our colleagues, by contrast, would have the district court ignore such an individualized determination, and instead assume that the offense occurred in a hypothetical average jurisdiction.

Our colleagues also argue that the added deterrence needs in Puerto Rico did not justify this sentence because the defendant had not yet shot anyone or "otherwise added to the violence in Puerto Rico." Rather, he was "only" found to have been in a McDonald's parking lot with a loaded machine gun. But if you want to deter machine gun violence, deterring possession of machine guns -- particularly by individuals like Flores with no weapons training -- is perfectly logical. The law, after all,

12

punishes both possession and use, and there is nothing in section 3553 that bars added deterrence aimed at both. And why was the defendant carrying a fully loaded gun if not anticipating a possible need to pull the trigger (and thereby generate even more spent shells)?

Finally, our colleagues suggest that the belief held by many district court judges who live in Puerto Rico that Puerto Rico experiences an atypically high level of machine gun violence was not "reasonable" or "reliabl[e]." This is a remarkable assertion, especially given that our court has itself previously held that "[t]he district court . . . did not err in considering the problem of gun violence in Puerto Rico and that 'Puerto Rico is a hot spot for weapons.'" Viloria-Sepulveda, 921 F.3d at 10; see also United States v. Delgado-Sánchez, 849 F.3d 1, 12–13 (1st Cir. 2017) ("[Defendant] claims that it was error for the district court to take note of Puerto Rico's significant problem with gun violence. Our precedent flatly rejects this argument."). In any event, this argument is unpreserved. At sentencing, the government expressly stated in its opening argument that Puerto Rico was a "hotspot for gun violence." Defense counsel (who also lives in Puerto Rico) voiced no disagreement at all. The district court itself then explained that "[t]he impact in Puerto Rico of this particular offense is more serious than that considered by the Sentencing Commission." Again, defense counsel raised no

13

objection to the factual assertion implicit in this statement. Flores cannot challenge the district court's factual description of conditions in Puerto Rico without at least carrying the burden of showing obvious error. See United States v. Rondón-García, 886 F.3d 14, 20 (1st Cir. 2018). To the extent that Flores now argues that the sentencing court procedurally erred because it based its assessment of community characteristics on its own perceptions without citing to data, he has "doubly waived this argument" because he did not raise it below or in his opening brief. See United States v. Leoner-Aguirre, 939 F.3d 310, 319 (1st Cir. 2019). Further, even in his belated challenge, Flores does not advance an argument that this was plain error under the applicable four-part test.3 In sum, we reject the limitations as fashioned and applied by our colleagues. We conclude, instead, that an increased need for deterrence in a given community may (but need not) justify a variant sentence.

3 "[A] reviewing court may set aside a challenged portion of a criminal sentence if, and only if, the appellant succeeds in showing (1) that an error occurred (2) which was clear or obvious and which not only (3) affected the defendant's substantial rights, but also (4) seriously impaired the fairness, integrity, or public reputation of judicial proceedings." United States v. Pabon, 819 F.3d 26, 33 (1st Cir. 2016) (alterations in original) (quoting United States v. Padilla, 415 F.3d 211, 218 (1st Cir. 2005)).

14

III.

None of this means that an upward variance based on community characteristics will always withstand a challenge on appeal. First, a defendant can insist that the community characteristics must be derived from information that is "reliable and accurate." United States v. Díaz-Rivera, 957 F.3d 20, 27 (1st Cir. 2020) (quoting United States v. Tavano, 12 F.3d 301, 305 (1st Cir. 1993)). Second, the community characteristics must be relevant to the individual defendant or the charged offense. To put a finer point on it, when the charged offense is possession of a machine gun, community characteristics regarding the prevalence of gun violence in the community in which the offense took place is sufficiently relevant. Third, the sentencing court must consider all relevant section 3553(a) factors. Finally, any variance imposed must be substantively reasonable; it must have both "a plausible sentencing rationale and a defensible result." United States v. Contreras-Delgado, 913 F.3d 232, 243 (1st Cir. 2019) (quoting United States v. Zapata-Vázquez, 778 F.3d 21, 24 (1st Cir. 2015)). In this way, the reliance on community characteristics cannot be "too much."

The eighteen-month upward variance imposed in this case is reasonable under this analysis. During sentencing, the government stressed -- without objection or challenge -- that "Puerto Rico is a hot-spot of gun violence." The sentencing judge

15

concurred with that statement, opining that "crime in Puerto Rico far exceeds the known limits on the mainland. Even the Circuit Court of Appeals has recognized that." Indeed we have. See, e.g., United States v. Narváez-Soto, 773 F.3d 282, 286 (1st Cir. 2014) ("[Violent] crime is a real problem in Puerto Rico."); Zapata- Vázquez, 778 F.3d at 23 (firearm offenses are pervasive in Puerto Rico).

Flores possessed the loaded machine gun, additional ammunition, and an empty shell casing at the time of his arrest in Puerto Rico. As the district court observed, he had no training in the safe use of the gun, and did not appear to have the means to have purchased the gun. The court considered Flores's acceptance of responsibility, lack of prior criminal convictions, age, education, circumstances of his arrest for the charged conduct, and the high rate of violent crime in Puerto Rico and highly dangerous nature of machine guns. The court confirmed it had "considered the other sentencing factors set forth in Title 18, United States Code section 3553(a)." Finally, the district court's rationale for imposing the upward variance -- "reflect[ing] the seriousness of the offense, promot[ing] respect for the law, protect[ing] the public from further crimes by Mr. Flores, [and] address[ing] the issues of deterrence and punishment" -- is plausible. These concerns are directly related to the court's clearly articulated observation about the need to

16

deter violent crime in Puerto Rico. The resulting upward variance of eighteen months, although certainly large, is defensible. The resulting sentence is one-third the ten-year statutory maximum. See 18 U.S.C. § 924(a)(2). And looking at the record as a whole and the district court's clear explanation, "we cannot say that [this] sentence, though upwardly variant, falls outside the broad universe of reasonable sentences." United States v. Vélez-Andino, 12 F.4th 105, 117 (1st Cir. 2021). See United States v. Rivera- Morales, 961 F.3d 1, 20 (1st Cir. 2020) ("[T]here is no one reasonable sentence in any given case but, rather, a universe of reasonable sentencing outcomes." (quoting United States v. Clogston, 662 F.3d 558, 592 (1st Cir. 2011))).

IV.

Given the unfortunate 3-3 split of our court in this case, it is fair to ask, "what next?" First, the sentence in this case is affirmed. Savard, 338 F.3d at 25. Second, Carrisquillo- Sanchez, Rivera-Berrios and Flores-Machicote remain controlling circuit precedent unless and until a majority in an en banc hearing or the Supreme Court rules otherwise. Third, whether an upward variance based on a higher than average rate of gun violence in a community can be justified as a Kimbrough policy disagreement remains unresolved.

17

THOMPSON, Circuit Judge, with whom BARRON, Chief Judge, and MONTECALVO, Circuit Judge, join. The result of today's en banc decision is that Emmanuel Flores-González's upwardly variant sentence still stands, even though cases requiring us to vacate his sentence remain good law — and so (obviously) continue to bind future panels of this Circuit and the district courts within it as well.

This pitch-perfect quote by the Supreme Court helps set the stage (no need to memorize all that we say in this set-up section, because we will resay and explain things as we go along):

It has been uniform and constant in the federal judicial tradition for the sentencing judge to consider every convicted person as an individual and every case as a unique study in the human failings that sometimes mitigate, sometimes magnify, the crime and the punishment to ensue.

Gall v. United States, 552 U.S. 38, 52 (2007) (emphases added)

(quoting Koon v. United States, 518 U.S. 81, 113 (1996)).

No one doubts that federal sentencing is a tough task, especially in the world of the now-advisory sentencing guidelines — a manual running many hundreds of pages that district judges use to orient their thinking. See Molina-Martinez v. United States, 578 U.S. 189, 193 (2016). A lot is on the line "for the defendant, the victim, and the community." See United States v. Sabillon- Umana, 772 F.3d 1328, 1330 (10th Cir. 2014) (Gorsuch, J., for the court). Using suitably reliable information, judges (speaking

18

generally, and per the legal commands in play here) must craft "individualized" sentences — whether within or without the proposed guidelines range, though still respecting statutory minimums and maximums — that are no greater than necessary to satisfy approved goals like rehabilitation, public protection, and deterrence (a guidelines range, incidentally, is the sentencing commission's estimate of a reasonable range of punishment for "each category of offense involving each category of defendant," see 28 U.S.C. § 994(b)).4 See Gall, 552 U.S. at 50. The takeaway catchphrase then is that "[f]air sentencing is individualized sentencing." See U.S. Sentencing Comm'n, Fifteen Years of Guidelines Sentencing: An Assessment of How Well the Federal Criminal Justice System is Achieving the Goals of Sentencing Reform 113 (2004) (emphasis added).

While easy to state these rules (as just indicated) can be hard to apply. But apply them we should — indeed must — in this appeal by Flores (as we will call him, consistent with Spanish naming customs).

As a refresher on the facts, a then 19-year-old Flores pled guilty a few years back to unlawfully possessing — but not illegally using — a Glock pistol altered to fire as a machine gun. He had no criminal priors. The district judge calculated his

4Because no statutory maximum or minimum rears its head here, our focus is on sentencing outside those worlds.

19

advisory prison range as 24 to 30 months. And Flores and the government recommended sentences within that range. But the judge picked 48 months — 100% above the bottom of the range, 60% above the top of the range.

Nothing about Flores's own past conduct or the individual way he committed the crime — other than his having committed it in Puerto Rico — drove the judge's sizable upward variance. And this we know from the judge's own words. "The [c]ourt," said the judge, did "not purport to establish that . . . Flores'[s] crime itself was more harmful than others similar to his." Rather, the judge explained, what triggered the major variance was that Flores's crime fell "within a category of offenses, gun crimes, that the [c]ourt, considering the particular situation in Puerto Rico [involving violence], views as more serious here than if they had occurred in a less violent society." And before revealing Flores's sentence, the judge played an audio and video recording of a "recent" machine-gun "massacre" that even he agreed had no relation to Flores's own specific conduct apart from his having illegally possessed the gun in Puerto Rico.

Relying on United States v. Rivera-Berríos, 968 F.3d 130 (1st Cir. 2020), and United States v. Carrasquillo-Sánchez, 9 F.4th 56 (1st Cir. 2021), a panel of our court vacated Flores's sentence as procedurally unreasonable. See United States v. Flores- González, 34 F.4th 103, 118 (1st Cir.), withdrawn on grant of reh'g

20

en banc, 46 F.4th 57 (1st Cir. 2022). To oversimplify slightly, the panel so ruled because the judge based the upward variance solely on the community characteristics of the crime's locale — without connecting his decision to "a 'special characteristic attributable either to the offender' or the circumstances of 'the offense.'" See Flores-González, 34 F.4th at 118 (quoting Rivera- Berríos, 968 F.3d at 137). A concurring panelist — the author of our colleagues' opinion (the opinion appearing before the one you're reading now) — "agree[d] that our most recent precedent under Rivera-Berríos and Carrasquillo-Sánchez precludes us from affirming." See id. at 121 (Kayatta, J., concurring) (emphases added). But he thought that those two decisions wrongly strayed from the lane identified in United States v. Flores-Machicote, 706 F.3d 16 (1st Cir. 2013), see Flores-González, 34 F.4th at 119 (Kayatta, J., concurring) — a decade-old opinion that lets judges impose upwardly variant sentences based on community characteristics, so long as they do not go "too far" by focusing "too much on the community and too little on the individual," see Flores-Machicote, 706 F.3d at 24.

Similarly convinced (though offering a different metaphor) that Rivera-Berríos injected "error into this [c]ourt's caselaw that has since metastasized," the government asked us to cure that perceived flaw through en banc review.

21

A majority of then-active judges voted to rehear the case en banc and vacated the panel's opinion. The en banc court now divides evenly on whether the district judge erred or not in varying upward, which leaves the caselaw compelling vacatur untouched (and binding on future panels) but affirms the district court's judgment by operation of law — thus dashing Flores's hopes for a new sentence.

Our colleagues' opinion defends the district judge's action as within his discretion. But still-governing precedent dictates the opposite conclusion, as we will explain in the many pages that follow.

I

We begin by briefly revisiting the central facts and prior proceedings.5

A

Puerto Rico police agents got an arrest warrant for Flores on domestic violence and weapons charges. Having heard that he would be at a local McDonald's, they stopped him after he went through the eatery's drive-thru. Arresting him, they found a Glock pistol modified to fire automatically, 63 rounds of

5 Because this appeal follows a guilty plea, we draw the background information from the probation office's presentence report and the transcripts of the relevant court proceedings. See, e.g., United States v. Edwards, 857 F.3d 420, 421 n.1 (1st Cir. 2017).

22

ammunition, and a spent shell casing (among other items). And this incident led to his being charged federally with unlawfully possessing a machine gun, to which he pled guilty. See 18 U.S.C. § 922(o).

B

We skip straight to sentencing. Adopting probation's presentence report, the judge (over the defense's objection) set Flores's base offense level at 20 for possessing the machine gun as a "prohibited person" because of his self-admitted drug use, see USSG § 2K2.1(a)(4)(B), but subtracted 3 levels because of his acceptance of responsibility, see USSG § 3E1.1(a) — for a total offense level of 17. The judge then pegged Flores's criminal history category at I, the lowest category (the presentence report noted that the commonwealth court dismissed the domestic-violence charges at the preliminary-hearing stage). This left Flores with an advisory prison range of 24 to 30 months. The defense requested 24 months. The government proposed 30 months.

Directing the parties' attention to the "sentencing factors" in 18 U.S.C. § 3553(a), the judge said that gun-related "crime in Puerto Rico far exceeds the known limits on the mainland" and that "[v]iolent crime and murders are occurring at all hours of the day, in any place on the island" — "even on congested public highways, in shopping centers, public basketball courts, and at cultural events." And the judge added that "[m]achine guns, like

23

the one . . . Flores possessed in this case, are present everywhere, obtained by persons, like . . . Flores, who have no training in the proper use of weapons and who appear not to have the means to purchase them."

Continuing, the judge stated that he had considered "the community and geographic factors" like the "high firearms and violent crime rate" — with "the community" here being "the entire island . . . because weapons crimes are not limited to one particular area or region." Also believing — without offering any reliably confirming evidence — that "[t]he impact in Puerto Rico of this particular offense is more serious than that considered by the [s]entencing [c]ommission when it drafted the guidelines," the judge then noted that "[d]eterrence" — stopping "criminal behavior by the population at large" — "is an important factor" in the § 3553(a) "calculus."

The judge did mention some biographical information (Flores's earning a "GED" certificate, his "history of using marijuana," and his Puerto Rico "arrest") and recounted some details about the offense (the police's confiscating the Glock, the 63 rounds of ammo, and the spent casing). But the judge specifically and clearly said that he "d[id] not purport to establish that . . . Flores's crime itself was more harmful than others similar to [Flores's]." On the contrary, the judge — in keeping with his theme — said that Flores's crime came "within a

24

category of offenses, gun crimes, that the [c]ourt[] [(that is, the judge,)] considering the particular situation in Puerto Rico, views as more serious . . . than if they had occurred in a less violent society."

Still focused on Flores's machine-gun "possess[ion],"

the judge remarked that a weapon like that "can fire more than a thousand rounds per minute which allows a shooter to kill dozens of people within a matter of seconds." Aside from "bombs, missiles, and biochemical agents," the judge could "conceive of few weapons that are more dangerous than machine guns." Machine guns are "unusual," the judge said, and beyond "a few [g]overnment- related uses, [they] largely exist on the black market." Wanting to show "[t]he dangerousness of a machine gun," the judge then played an audio and video recording of a "recent massacre" at a Puerto Rico housing project "where six persons were machine-gunned to death in a matter of seconds." "This," the judge said, "is what some people in Puerto Rico live with every single day."

Convinced that neither party's proposed sentence "reflect[ed] the seriousness of the offense, promote[d] respect for the law, protect[ed] the public from further crimes by . . . Flores," or "address[ed] the issues of deterrence and punishment," the judge imposed a variant sentence of 48 months — 18 months more than the top of the recommended sentencing range.

25

Asked by the judge if there was "[a]nything else,"

defense counsel objected to the variant sentence as both "procedurally and substantively unreasonable." Counsel disagreed with the judge's view that the "guideline[s] do[] not adequately reflect the possession of a machine gun when determining the applicable" sentencing range and thought that the judge had not "adequately consider[ed]" the proper sentencing factors. Counsel also objected to the use of the audio and video of the machine- gun massacre. "That video," said counsel, "is completely unrelated to the facts of this case, does not reflect . . . Flores'[s] conduct in this case, and —" at which point the judge interrupted, saying: "It's not supposed to. It's just supposed to show what a machine gun can do." Counsel responded that he did not believe "it was necessary" to "play[] . . . the video and audio of that shooting to demonstrate it when it's already considered within the guidelines." And counsel called the sentence "substantively unreasonable" given Flores's "characteristics" and status as "a first-time offender." But the judge would have none of it.

The post-sentencing "Statement of Reasons" form — which sentencers complete under 18 U.S.C. § 3553(c)(2) — included the judge's comment that "the impact of this [kind of weapon] on the Island is more serious tha[n] that considered by the [s]entencing

26

[c]ommission."6 And under the heading "18 U.S.C. § 3553(a) and other reason(s) for a variance (Check all that apply)," three boxes were checked: to protect the public, to deter others from copying the crime, and to reflect how serious the crime was. Among the boxes left unchecked was one labeled "Policy Disagreement with the Guidelines (Kimbrough v. U.S., 552 U.S. 85 (2007)[)]."7

C

From this sentence Flores appealed, raising three main issues. He first argued that the judge procedurally erred by labeling him a "prohibited person." He next argued that the judge procedurally erred by giving too much weight to community factors and failing to individually design his sentence. And he last

6 Section 3553(c)(2) pertinently says that a judge picking a sentence "outside the [guidelines] range" must provide the reasons for the pick "with specificity in a statement of reasons form." The commission uses the information in these documents "to make recommendations to Congress." See United States v. Morales- Negrón, 974 F.3d 63, 68 (1st Cir. 2020). See generally United States v. Murchison, 865 F.3d 23, 26-27 (1st Cir. 2017) (mentioning that the bureau of prisons also uses this data to "classif[y] and process[] sentenced offenders"). A standing order of the district court provides that probation shall fill out these forms based on the judges' in-court comments and send them to the judges for final approval, see Morales-Negrón, 974 F.3d at 68 — apparently as a way to "streamline" the process, see Standing Order No. 17-205 (Apr. 28, 2017) (adding as well that judges "shall" give the parties sealed copies of these documents on request).

7 Granting Flores's request for access to the statement of reasons, the judge's electronic order says that "the transcript of the sentencing hearing is the official document and sets forth the [c]ourt's reasoning for the sentence imposed." But the fact remains that the judge left the box blank, despite having had the opportunity to check it.

27

argued that the judge substantively erred by imposing such a stiff sentence on a first-time offender for no other reason than that he had a machine gun. The panel affirmed on the first issue, reversed on the second, and did not reach the third. The full court then granted rehearing en banc, vacating that opinion — a result that required us to reassess the case from scratch.

II

Before addressing Flores's particular challenges, we provide some context about the individualized-sentencing requirement itself — as often "a page of history is worth a volume of logic." See N.Y. Trust Co. v. Eisner, 256 U.S. 345, 349 (1921) (Holmes, J., for the Court). So we start with a summary of federal sentencing, explaining along the way (among other things) how it went "from total judicial discretion, to virtually none with mandatory guidelines, and back to advisory guidelines with discretion for variances and policy disagreements with the guidelines." See Mark W. Bennett, Addicted to Incarceration: A Federal Judge Reveals Shocking Truths About Federal Sentencing and Fleeting Hopes for Reform, 87 UMKC L. Rev. 3, 22 (2018).8 Excuse us if we run a little long here. But the parties and our colleagues' opinion present a variety of arguments about

8 Former Judge Bennett was a district judge in the Northern District of Iowa from 1994 to 2019. See Biographical Directory of Article III Federal Judges — Bennett, Mark W., Federal Judicial Center, https://www.fjc.gov/history/judges/bennett-mark-w. A

28

the panel decision's approach under the current advisory sentencing regime. And so understanding what we and our judicial superiors have and have not said about sentencing will put the reader in the right frame of mind for the analysis to come. To give a sneak-peek preview, what follows will make clear that judges must "individualize" sentences after the Sentencing Reform Act of 1984 (which created the sentencing commission and authorized the sentencing-guidelines system), just as they had to do before that statute's enactment. And so it will bring into sharp focus the question whether an upwardly variant sentence for a machine-gun- possession offense is "individualized" when based solely on the level of violent crime in the geographic community where the offense occurred.9

A

1

In the early(ish) days of the Republic, Congress gave federal judges enormous sentencing discretion. See, e.g., Kate Stith & Steve Y. Koh, The Politics of Sentencing Reform: The Legislative History of the Federal Sentencing Guidelines, 28 Wake

year before he left the bench, Bennett noted that he had "sentenced more than 4,000 offenders to federal prison." See Addicted to Incarceration, 87 UMKC L. Rev. at 3.

9Note too that because no summary can include every detail and stay a summary, we do not discuss all the jots and tittles of federal sentencing.

29

Forest L. Rev. 223, 225 (1993). These judges, "in most cases, could sentence anywhere from probation to the statutory maximum sentence and there was little appellate review" of their decisions. Addicted to Incarceration, 87 UMKC L. Rev. at 7 (emphasis added); see also The Politics of Sentencing Reform, 28 Wake Forest L. Rev. at 226 (stating that "[f]or over two hundred years, there was virtually no appellate review of the trial judge's exercise of sentencing discretion"). So in this time of "sweeping" sentencing authority — described by one respected jurist (the late Judge Marvin Frankel, the "father of sentencing reform") as "terrifying and intolerable for a society that professes devotion to the rule of law" — if an appellate court concluded that the judge imposed a sentence "within statutory limits," the ruling was "generally not subject to review." See The Politics of Sentencing Reform, 28 Wake Forest L. Rev. at 228 (first three quotes (citations omitted)); United States v. Tucker, 404 U.S. 443, 447 (1972) (fourth and fifth quotes).10

We have relied on Judge Frankel's writings before.

10 See United States v. Foss, 501 F.2d 522, 527 (1st Cir. 1974) (citing Marvin E. Frankel, Criminal Sentences: Law Without Order (1972)). After stints as a litigator and scholar, he served as a district judge in the Southern District of New York from 1965 to 1978. See Biographical Directory of Article III Federal Judges — Frankel, Marvin E., Federal Judicial Center, https://www.fjc.gov/history/judges/frankel-marvin-e. Among other points, he argued passionately that

[c]orrectly understood, the "discretion"

of judicial officers in our system is not a

30

Even in that era of wide-ranging sentencing discretion, reviewing courts did vacate sentences when judges failed to individualize them. A case in point is United States v. Wardlaw, decided before the passage of the Sentencing Reform Act and holding that the district court there had "exceeded the bounds of its sentencing discretion" by not "individualiz[ing]" the prison terms. See 576 F.2d 932, 938 (1st Cir. 1978).

Troubled by the drug dealing in Puerto Rico, the district court sentenced two "mules" (drug carriers) "harsh[ly]" (though less than the statutory maximums) so "word [will] spread around that these mules are getting worse treatment than the mule owners and then they will stop being mules because they have to think about it twice before they proceed and no one is going to do the dirty work for [the owners]." Id. at 936. Agreeing that a district court's "duty to take the individual defendant into account did not mean [the lower court] could not assess the sentence's presumed

blank check for arbitrary fiat. It is an authority, within the law, to weigh and appraise diverse factors (lawfully knowable factors) and make a responsible judgment, undoubtedly with a measure of latitude and finality varying according to the nature and scope of the discretion conferred. But "discretionary" does not mean "unappealable."

Discretion may be abused, and discretionary decisions may be reversed for abuse.

See Criminal Sentences at 84.

31

effects on others, and that general deterrence was a legitimate factor to be considered in arriving at a sentence," we added a "but" — "[b]ut always these effects had to be considered along with the individual circumstances of the defendant." Id. at 938 (emphases added).11 "The court's duty to 'individualize' the sentence," we continued,

simply means that, whatever the judge's thoughts as to the deterrent value of a jail sentence, he must in every case reexamine and measure that view against the relevant facts and other important goals . . . . Having done so, the . . . judge must finally decide what factors, or mix of factors, carry the day.

While the judge's conclusions as to deterrence may never be so unbending as to forbid relaxation in an appropriate case, they may nonetheless on occasion justify confinement although other factors point in another direction.

Id. (quoting Foss, 501 F.2d at 528). And we ultimately ruled that the "usual individual considerations" — "mitigating factors" like "defendants' youth, positive presentence reports, and lack of criminal records, and even such aggravating factors" like "the large amount of cocaine involved" — "played little or no part in [the judge's] thinking." Id. at 938-39 (emphasis added). Mincing no words, we said that "mechanistic" sentencing that steadfastly

11 Deterrence comes in two forms — specific (deterring the defendant from committing crimes after release) and general (deterring others from heading down the same criminal path).

32

snubs "individual" differences will not do. See id. at 938 (quoting Foss, 501 F.2d at 527).

Also error was how the judge "focused to the exclusion . . . of all else on the assumed impact upon the large dealers who 'run' the smugglers of meting out inflexibly harsh sentences to their agents." Id. at 939 (emphasis added). Though "judges have considerable discretion in sentencing," we held that they may neither "relentlessly pursue at a defendant's cost a single, questionable theory while simply brushing aside all the other criteria commonly weighed by the vast majority of sentencing courts," nor use sentences "chiefly as instruments of retaliation against other, different criminals." Id. (emphasis added).

And during this era of wide-open discretionary sentencing, we were not alone in occasionally vacating sentences because a judge flouted the individualized-sentencing rule by overly focusing on the perceived need for general deterrence. Other courts did too, like when a judge had a policy of always giving maximum sentences for certain crimes. See, e.g., Foss, 501 F.2d at 527 (citing cases from the Third, Fifth, Sixth, and D.C. Circuits).

B

1

Responding (eventually) to reformers' complaints that federal sentencing (as it then existed) was really "a non-system

33

in which every [district] judge is a law unto himself or herself," with a defendant's sentence turning "on the judge he or she gets," see Marvin E. Frankel, Jail Sentence Reform, N.Y. Times, Jan. 15, 1978, at E21, Congress passed the Sentencing Reform Act of 1984. This law aimed to curb "variable sentencing caused by different judges' perceptions of the same criminal conduct," see United States v. Kirkpatrick, 589 F.3d 414, 416 (7th Cir. 2009) (Easterbrook, C.J., for the court), promoting sentencing uniformity while also ensuring that sentencing stayed individualized and so tailored to the offender and the specific offense committed.

To that end, the Sentencing Reform Act set up the sentencing commission (just "commission" from here on), a non- elected body within the judicial branch that developed a sentencing-guidelines regime that remains in place today. See Mistretta v. United States, 488 U.S. 361, 367-68 (1989). With these changes, Congress sought (among other things) to channel sentencers' discretion and reduce sentencing disparities — a technical phrase describing the indefensibly wide range of punishments different judges had imposed on similarly situated defendants (thus raising equal-protection concerns in the minds of some reformers). See id. at 366-67; Michael M. O'Hear, The Original Intent of Uniformity in Federal Sentencing, 74 U. Cin. L. Rev. 749, 761 (2006).

34

"The federal sentencing guidelines are" (as one commentator pithily put it) "simply a long set of instructions for one chart: the sentencing table." Frank O. Bowman III, The Failure of the Federal Sentencing Guidelines: A Structural Analysis, 105 Colum. L. Rev. 1315, 1324 (2005). Roughly speaking, a judge scores the crime's base offense level, adjusting for certain aggravating or mitigating circumstances to get the total offense level. See USSG § 1B1.1. Next the judge scores the defendant's criminal record to get the criminal history category. See id. Turning then to the guidelines' sentencing table, the judge marks — with fingers (for example) — the total offense level on the table's vertical line and the criminal history category on the horizontal line. And "[w]here the judge's finger[s] stop[], he or she finds" the sentencing range inside the statutory minimums and maximums. See Albert W. Alschuler, The Failure of the Sentencing Guidelines: A Plea for Less Aggregation, 58 U. Chi. L. Rev. 901, 907 (1991).

For years judges working under the Sentencing Reform Act generally had to pick from the guidelines range. See Koon, 518 U.S. at 108. Free-wheeling discretion was out, though judges were guided somewhat by factors in § 3553(a) — of which there were and are seven:

Factor one is "the nature and circumstances of the offense and the history

35

and characteristics of the defendant." 18 U.S.C. § 3553(a)(1). Factor two is

the need for the sentence . .

. (A) to reflect the seriousness of the offense, to promote respect for the law, and to provide just punishment for the offense; (B) to afford adequate deterrence to criminal conduct; (C) to protect the public from further crimes of the defendant; and (D) to provide the defendant with needed educational or vocational training, medical care, or other correctional treatment in the most effective manner.

Id. § 3553(a)(2). Factor three is "the kinds of sentences available." Id.

§ 3553(a)(3). Factor four is the guidelines. Id. § 3553(a)(4). Factor five is "any pertinent policy statement . . . issued by the [s]entencing [c]ommission." Id.

§ 3553(a)(5). Factor six is "the need to avoid unwarranted sentence disparities."

Id. § 3553(a)(6). And factor seven is "the need to provide restitution to any victims."

Id. § 3553(a)(7).

United States v. Correa-Osorio, 784 F.3d 11, 28 n.24 (1st Cir. 2015); see also Rita v. United States, 551 U.S. 338, 347-48 (2007).

Judges in this era could sentence above or below the guidelines range in specific cases (known as departures, in law- speak) — but only if "there exist[ed] an aggravating or mitigating circumstance of a kind, or to a degree, not adequately taken into consideration by the . . . [c]ommission in formulating the guidelines[.]" See Koon, 518 U.S. at 92 (quotation marks omitted).

36

The commission, after all, sets the range with the typical case in mind — thus ensuring that an offender gets sentenced both individually and comparably to others who did the same crime (when nothing about the offender or the crime's commission made either relevantly different from other perpetrators of the same crime). And because judges then had to "sentence within the applicable [g]uidelines range (in the absence of circumstances that justify a departure)," see United States v. Booker, 543 U.S. 220, 259 (2005), "the uses that [they] could make of the factors listed in section 3553(a) were severely circumscribed . . . to preserve the mandatory character of the guidelines," see United States v. Dean, 414 F.3d 725, 728 (7th Cir. 2005) (Posner, J., for the court) — which is why we said "guided somewhat by the factors in § 3553(a)," several lines ago.

The Sentencing Reform Act also erected "standards of appellate review for certain claims of sentencing error." United States v. Pho, 433 F.3d 53, 60 (1st Cir. 2006), abrogated on other grounds by Kimbrough v. United States, 552 U.S. 85 (2007). One provision (for example) mandated de novo review of guidelines departures, see Booker, 543 U.S. at 261 — meaning reviewing courts acted without deference to the district judges' views, see Toddle Inn Franchising, LLC v. KPJ Assocs., LLC, 8 F.4th 56, 66 (1st Cir. 2021) (explaining what de novo review means).

37

The guidelines are no longer binding thanks to Booker, which ushered in the new age of federal sentencing (in place at the time of Flores's sentencing and at present).

Booker said that a mandatory sentencing regime based on judge-made findings violated the Sixth Amendment's jury-trial guarantee. See 543 U.S. at 243-45. And Booker fixed that fatal constitutional defect by erasing two parts of the Sentencing Reform Act going forward: the one that forced judges to impose sentences within the guidelines range, and the other that required de novo review of guidelines departures. See id. at 259.

But despite dropping the guidelines from rules to advice, Booker still told judges to consider the guidelines during sentencing and to base their outcomes on the § 3553(a) factors. Id. at 259-60. That directive gave those factors — "made dormant" by "the mandatory application of the [g]uidelines" — a brand-new "vitality." See United States v. Trujillo-Terrazas, 405 F.3d 814, 819 (10th Cir. 2005) (McConnell, J., for the court). See generally Scott Michelman & Jay Rorty, Doing Kimbrough Justice: Implementing Policy Disagreements with the Federal Sentencing Guidelines, 45 Suffolk U. L. Rev. 1083, 1095 (2012) (explaining that after Booker "[m]andatory [g]uideline[s] sentencing was out" and "[t]he seven"

38

§ 3553(a) factors "were in").12 And to fill the standard-of-review hole, Booker told appellate courts to inspect sentences only for "reasonableness." See 543 U.S. at 261.

3

Then came Rita.

Fleshing out the advisory-guidelines regime, Rita said that because the guidelines reflect the commission's bid to reconcile the § 3553(a) factors in the typical case, so should the judges' sentencing decisions. See 551 U.S. at 347-48. Thus when sentences jibe with the guidelines' application of those factors in the "mine run" of cases, they are "probabl[y]" reasonable (and reviewing courts "may" presume that within-guidelines sentences are reasonable). See id. at 351. But if they do not jibe with

Guidelines-based departures still exist after Booker. See, 12

e.g., Irizarry v. United States, 553 U.S. 708, 714 (2008). Although they lead to the same result (a sentence outside the guidelines range), variances and departures get there via different routes. See United States v. Fletcher, 56 F.4th 179, 187 (1st Cir. 2022). A variance is a non-guidelines sentence based on the judge's consideration of the § 3553(a) factors. See United States v. Vixamar, 679 F.3d 22, 33 (1st Cir. 2012). A departure, contrastingly, is a "non-[g]uidelines sentence imposed under the framework set out in the [g]uidelines" — including the departure provisions. See Irizarry, 553 U.S. at 714. See generally Fletcher, 56 F.4th at 187 (noting "that a departure is just a variance by another name" since "there is no departure that could not be justified as a variance," but adding that "we cannot entirely abandon the nomenclature" (quotation marks omitted)). A judge must give the parties "advance notice of the grounds for any contemplated departure." See Fletcher, 56 F.4th at 187-88. But the judge must "also avoid unfair surprise when adopting a variance." See id. at 188.

39

the guidelines, that same probability does not exist (though reviewing courts may not presume an outside-the-guidelines sentence is unreasonable). See id. at 354-55.

Judges, Rita added, must explain the reasons for their chosen sentences — i.e., they must say enough to show the appellate courts that they "considered the parties' arguments and ha[d] a reasoned basis for exercising [their] own legal decisionmaking authority." Id. at 356. Within-guidelines sentences do "not necessarily require lengthy explanation." Id. But sentences falling outside the guidelines require that judges "explain why" they decided not to follow the commission's recommendations. See id. at 357.

Finally and separately, Rita also made clear that "[a]ppellate 'reasonableness' review" translates to abuse-of- discretion review. See id. at 351.

4

Enter Gall, which further clarified the new advisory-

guidelines scheme.

Booker said that judges are not bound by the commission's suggested guidelines, even in the mine-run case. But Gall reminded judges that they must know what the suggestions are (and thus the ranges recommended for the defendants' sentences in the cases at hand) before using their discretion. See 552 U.S. at 49. So judges must first calculate the advisory ranges. See id. And

40

without "presum[ing]" that a guidelines sentence is reasonable, they must next make a particularized assessment (based on the facts presented) of the § 3553(a) concerns — considered the hallmark of "individualized" sentencing — and then explain the reasons for the selected sentences. See id. at 50; see also id. at 52.

And staying with explanations, Gall nixed any idea that judges must justify outside-guidelines sentences with "extraordinary" circumstances. See id. at 47. But judges must give "serious consideration to the extent of any" deviation "from the [g]uidelines" and offer "sufficient justifications" for unusually heavy or light sentences. See id. at 46. "[M]ajor" guidelines deviations "should be supported by a more significant justification than . . . minor one[s]," Gall noted. Id. at 50. Which means that any justification must be "sufficiently compelling to support the degree of the variance." See id. But in all cases judges must offer "adequate[]" explanations "to allow for meaningful appellate review and to promote the perception of fair sentencing." See id.

Gall also noted that circuit courts must review all sentences — whether within or without the guidelines — only for reasonableness under an abuse-of-discretion standard, see id. at 46, and that reasonableness has both procedural and substantive components, see id. at 51. A sentence is procedurally unreasonable if the judge "fail[ed] to calculate (or improperly calculat[ed])

41

the [g]uidelines range, treat[ed] the [g]uidelines as mandatory, fail[ed] to consider the § 3553(a) factors, select[ed] a sentence based on clearly erroneous facts, or fail[ed] to adequately explain the chosen sentence." Id. A sentence is substantively unreasonable if the judge acted too harshly given the "totality of the circumstances." See id.

5

On the very day Gall came out so too did Kimbrough, which highlighted another aspect of post-Booker sentencing discretion: judges can sentence outside the advisory-guidelines range not only (as Booker held) because of the nature and circumstances of the offense but also because of the nature of the guidelines themselves. See 552 U.S. at 91, 98, 108-10.

Dealing with a much-panned guidelines ratio that equated 1 gram of crack cocaine to 100 grams of powder cocaine, Kimbrough noted that judges and the commission have "discrete institutional strengths." See id. at 109. On the one hand, judges better know the particular defendants before them "than the [c]ommission or the appeals court[s]" and so are better positioned to apply the § 3553(a) considerations "in each particular case." Id. (quoting Rita, 551 U.S. at 357-58). On the other hand, the commission's expertise is mainly "empirical" — having as it does the knowledge, experience, and workforce to analyze data reflecting the combined experiences of sentencers across the country and the input of

42

different law-enforcement groups. See id. (quoting United States v. Pruitt, 502 F.3d 1154, 1171 (10th Cir. 2007) (McConnell, J., concurring)); see also Rita, 551 U.S. at 349.13 And if the commission settles on a policy choice for reasons beyond its expertise, the resulting guidelines may be attacked on that basis. See Kimbrough, 552 U.S. at 109.

These institutional differences also affect the degree of deference due a judge's decision to vary. In cases "outside" the guidelines' "heartland," decisions to vary "may attract greatest respect." See id. (quotation marks omitted). But in "mine-run" cases — average cases with no distinguishing circumstances — "closer review" may be necessary for decisions "based solely on" a policy disagreement with the guidelines. See id.

According to Kimbrough, the crack/powder ratio did not reflect the commission's usual method of using "empirical data and national experience." See id. The commission had cribbed it from the Anti-Drug Abuse Act, which represented Congress's assumptions

Because judges must take the guidelines "into account when 13

sentencing," see Booker, 543 U.S. at 264, the commission must continually update them to "encourag[e] . . . better sentencing practices" and "uniformity in the sentencing process," see id. at 263; see also Kimbrough, 552 U.S. at 108 (emphasizing that "Congress established the [c]ommission to formulate and constantly refine national sentencing standards"); Neal v. United States, 516 U.S. 284, 291 (1996) (stating that "Congress . . . charged the [c]ommission with the duty to measure and monitor the effectiveness of various sentencing, penal, and correctional practices").

43

about crack's dangerousness — assumptions later proved baseless and contrary to the commission's own research. See id. at 95-98, 109; see also id. at 111 (noting that the commission itself had taken the "consistent and emphatic position that the crack/powder disparity [was] at odds with § 3553(a)"). And adding everything up, Kimbrough held that because the ratio did not "exemplify the [c]ommission's exercise of its characteristic institutional role," judges abuse no discretion in ruling that that formula "yields a sentence 'greater than necessary' to achieve § 3553(a)'s purposes, even in a mine-run case," see id. at 109-10 — judges in other words can disagree with the guidelines (but not with statutes), though they must act reasonably in using that power.14

6

That is certainly a lot to take in. Which makes this a good time for a recap of certain facets of modern sentencing law that provide context for addressing the weight judges may give community characteristics in upwardly varying sentences.

14 When some post-Kimbrough opinions said that sentencers could vary from the guidelines only if the facts of the particular case made their application unjust, the Supreme Court replied that judges could "reject and vary categorically" from the at-issue guidelines "based on policy disagreement with them, and not simply based on an individualized determination that they yield an excessive sentence in a particular case." See Spears v. United States, 555 U.S. 261, 266 (2009) (first quote); id. at 264 (second quote).

44

Operating under the now-advisory-guidelines regime, and (importantly) considering each defendant as an individual, the judge starts by computing the relevant sentencing range — a product of the base offense level, any enhancing or mitigating adjustments to that level, the criminal history category, and any departures from the guidelines. See, e.g., United States v. Martínez-Benítez, 914 F.3d 1, 2 n.2 (1st Cir. 2019); United States v. Dávila- González, 595 F.3d 42, 46 (1st Cir. 2010). At that point the judge knows the commission-suggested range for the at-issue crime when committed by a defendant like the one before the court — i.e., a range missing any specific characteristic of the defendant or the way he committed the crime not captured by the guidelines themselves. And after letting the parties make their sentencing pitches, the judge then decides whether to vary from that range, using one or both of the following methods: categorically disagreeing with the suggested range — i.e., balking on a basis applicable to all defendants, "Eagle Scout[s]" and "street thug[s]" alike (for example), see United States v. Gully, 619 F. Supp. 2d 633, 643 (N.D. Iowa 2009); or by making an individualized appraisal of the § 3553(a) factors — i.e., by considering the particular characteristics of the defendant and the particular

45

offense conduct, see Kimbrough, 552 U.S. at 108-10; see also Gall, 552 U.S. at 51-52.

To achieve these ends — and also to satisfy the requirement "that 'the punishment should fit the offender and not merely the crime,'" see Pepper v. United States, 562 U.S. 476, 487-88 (2011) (quoting Williams v. New York, 337 U.S. 241, 247 (1949)) — the judge may consider any relevant evidence if it has sufficient signs of reliability and if the defendant had a chance to rebut it, see United States v. Hernández-Negrón, 21 F.4th 19, 25 (1st Cir. 2021) (declaring that due process requires that a judge not sentence a defendant on "false or materially incorrect" information). The judge next decides what term is appropriate (knowing that a sentence cannot be greater than necessary to satisfy federal-sentencing imperatives). See United States v. Jiménez-Beltre, 440 F.3d 514, 518-19 (1st Cir. 2006) (en banc). The judge then explains the sentence, including any detour from the range (the reasoning may sometimes — but not always — be apparent from context). See United States v. García-Carrasquillo, 483 F.3d 124, 132-33 (1st Cir. 2007). And while an "extraordinary" reason is not required for an outside-guidelines sentence, the reason must be "compelling" enough to justify the variance (ergo the bigger the variance the more robust the judge's explanation

46

must be, though there is no "rigid mathematical formula"). See Gall, 552 U.S. at 47, 50, 51.

We review all disputed sentences for reasonableness under an abuse-of-discretion standard. See Flores-Machicote, 706 F.3d at 20 (citing Gall, among other cases). And by playing our part (modest though it is) we ensure that sentencing practices stay fairly consistent, see Booker, 543 U.S. at 263, and that variances turn on proper considerations, see Rita, 551 U.S. at 382; see also Flores-Machicote, 706 F.3d at 20 — always remembering that a sentence must reflect an individualized assessment of the § 3553(a) factors through the lens of the particular defendant's case, see Nelson v. United States, 555 U.S. 350, 351 (2009) (per curiam) (stressing that the judge "must first calculate the [g]uidelines range, and then consider what sentence is appropriate for the individual defendant in light of the statutory sentencing factors, 18 U.S.C. § 3553(a), explaining any variance from the former with reference to the latter"); see also Booker, 543 U.S. at 261 (noting that "those factors in turn will guide appellate courts . . . in determining whether a sentence is unreasonable").

C

With that and at long last, we return to Flores's case.

In the pages to come we consider his challenge to the judge's prohibited-person ruling, which does not implicate the history just recounted about the individualized-sentencing requirement.

47

We then turn to his challenge that does implicate that history, which targets the judge's reliance on community characteristics and focuses on the many cases by us about whether and how those characteristics inform sentencing under the Sentencing Reform Act and the Supreme Court decisions interpreting it.

III

First up is Flores's claim that the judge's prohibited-

person finding constituted a procedural error.

A

As Flores rightly notes, a prohibited person in this context is someone "who engages in . . . regular use" of drugs "over a long period . . . proximate to or contemporaneous with the possession of the firearm." See United States v. Caparotta, 676 F.3d 213, 216 (1st Cir. 2012) (quotation marks omitted). Focusing on the "long period" language, he argues that because he admitted only to "a few months of drug use," his situation does not fit this definition. The government counters that the record readily supports the judge's finding.

We — as well as our colleagues in their opinion — think the government has the better of the argument.

B

During pretrial interviews Flores — who was 19 when nabbed — copped to smoking 4 or 5 marijuana joints daily since he

48

was 17 and to having smoked before his arrest.15 On the eve of sentencing, however, he claimed in a presentence interview that he only smoked regularly during the three months before his arrest. Probation suggested that Flores did this about-face only because he now realized that he could get a prohibited-person increase to his sentence. The judge accepted probation's position, thus triggering clear-error review. For our part, even assuming — without granting — that using marijuana for three months is not enough for prohibited-person status and that the late-in-the-game comment about the three months of marijuana use turns on a plausible view of the record, we think the judge's view is not implausible given Flores's earlier admissions about toking daily for years. See United States v. Marceau, 554 F.3d 24, 30-31 (1st Cir. 2009) (upholding a prohibited-person increase where, "even after [a] stay at a drug treatment facility," the defendant "was unable to remain drug-free" and where he admitted to "smok[ing] marijuana daily in the days before" his crime). And if "there is more than one plausible view of the circumstances, the [judge's] choice among supportable alternatives cannot be clearly erroneous." See United States v. Dunston, 851 F.3d 91, 101-02

15 Despite being legal for certain purposes in some states, marijuana remains illegal under federal law. See United States v. Ford, 625 F. App'x 4, 7 (1st Cir. 2015).

49

(1st Cir. 2017) (quoting United States v. Ruiz, 905 F.2d 499, 508 (1st Cir. 1990)).

IV

Now for the main event: Flores's claim that the sentencing judge procedurally erred by giving undue weight to the frequency of gun violence in Puerto Rico and thus not individually tailoring his sentence. Before taking that up directly, we review our precedent bearing on it — with us occasionally commenting on how this caselaw affects our colleagues' opinion. We then circle back to respond to the government's and our colleagues' opinion's specific arguments.

A

Our first case to substantially address whether and when community characteristics may guide sentencing under the Sentencing Reform Act is Flores-Machicote — a felon-gun-possession case that the government talks up in asking us to affirm Flores's sentence. Flores-Machicote rejected a defendant's claim that his upwardly variant sentence had to go because (per the defendant) the judge failed to treat him as an individual by fixating on geographic-based concerns — most notably and relevantly, "the incidence and trend lines of particular types of crime in the affected community." 706 F.3d at 23.

Flores-Machicote relied on our decision in United States v. Politano — an illegal-firearms-dealing case holding that judges

50

can "take into account all of the circumstances under which [the defendant] committed the offense, including the particular community in which the offense arose." See 522 F.3d 69, 74 (1st Cir. 2008) (explaining that "the district court expressly considered the ways in which Politano's firearms offense was more serious and harmful within this specific community"). But the judge there did not base the upwardly variant 24-month sentence (which exceeded the top end of the range by 6 months) solely on community characteristics and so without regard to the defendant's specific case. See id. Rather the judge gave individualized consideration to the defendant's situation beyond the offense's locale, noting the defendant's "likelihood of recidivism" — which the "[g]uidelines somewhat underestimate[d] or undercount[ed]" in his case. Id. And Politano did not itself say much about when if ever — outside the context of that specific case — community characteristics could drive a variant sentence. See id.

Flores-Machicote thus marked the first time a panel in this circuit developed a framework for analysis. Flores-Machicote (for example) explained that a community characteristic can "appropriately inform[] and contextualize[] the relevant need for deterrence," 706 F.3d at 23, which is a designated § 3553(a) factor. And Flores-Machicote said that a judge need not give each § 3553(a) factor equal prominence and so may give deterrence

51

greater weight than other factors more focused on the defendant's specific conduct or characteristics. See id.

But then (almost in the same breath) Flores-Machicote held that while judges can rely on geographic-specific concerns "not specifically tied to either the offender or the offense of conviction" — like the amount of gun crimes in a broad community — they cannot "go too far" in emphasizing those points. Id. at 24 (emphasis added). To help clarify the "go too far" concern, Flores-Machicote noted that a "judge's resort to community-based characteristics does not relieve him or her of the obligation to ground sentencing determinations in case-specific factors." See id. (emphases added and citing Politano, 522 F.3d at 74). Which is another way of saying that a judge must not "focus too much on the community and too little on the individual." See id. And when it comes to variances — even ones based in part on community characteristics — Flores-Machicote made sure to state that the judge "should typically . . . root[]" the reasons "either in the nature and circumstances of the offense or the characteristics of the offender." See id. at 21 (quotation marks omitted).

Moving then from the general to the concrete, Flores-

Machicote held that by "direct[ing] individualized attention to the defendant's case" — the "likely recidivism" of this particular offender played a major role in the sentencing decision (as the explanation showed) — the judge "did not cross this line" by

52

drawing on Puerto Rico's violent-crime rate in picking the sentence. See id. at 24 (emphasis added). And so the judge's choice survived abuse-of-discretion review. See id.

In letting judges rely on community characteristics — at least to some extent — in varying upward, Flores-Machicote tracked this Circuit's longstanding rule that a sentence must be "individualize[d]" rather than "mechanistic" and so cannot rest on a "questionable theory" of general deterrence to the exclusion of all else. See Wardlaw, 576 F.2d at 938-39.

And Flores-Machicote is not a one-off in letting community-crime-rate concerns "inform[] and contextualize[] the relevant need for deterrence" in a specific case. See 706 F.3d at 24 (emphasis added). As the government and our colleagues' opinion note, First Circuit caselaw is filled with opinions affirming upwardly variant gun-case sentences — with community-related factors appearing front and center in the judges' decisions.

But (a big "but" actually) these decisions (at least inferentially) — while noting that sentencers can consider a locale's violent crime rate — also noted (and crucially so!) that the judges there did explore the defendants' individual circumstances and "did not centrally rely on community considerations." See United States v. Robles-Pabon, 892 F.3d 64,

53

66 (1st Cir. 2018) (emphasis added).16 Our colleagues' opinion just ignores this aspect of those cases, including ignoring how at least one decision went out of its way to call Flores-Machicote's okaying reliance on community concerns "a limited grant of authority" that a sentencer cannot "stray beyond." See Bermúdez- Meléndez, 827 F.3d at 166 (emphasis added).

A good example of the many cases that spell trouble for the government and our colleagues' opinion is García-Mojica. The defendant there appealed a sentence of 100 months (8.33 years), an

16 To save space, we put the case cites here — listing them from oldest to newest: United States v. Naváez-Soto, 773 F.3d 282, 284, 286-87 (1st Cir. 2014); United States v. Santiago-Rivera, 744 F.3d 229, 231, 232-33 (1st Cir. 2014); United States v. Zapata- Vázquez, 778 F.3d 21, 22-23 (1st Cir. 2015); United States v. Díaz- Arroyo, 797 F.3d 125, 128-30 (1st Cir. 2015); United States v. Paulino-Guzman, 807 F.3d 447, 449, 451 (1st Cir. 2015); United States v. Pantojas-Cruz, 800 F.3d 54, 57, 59-60 (1st Cir. 2015); United States v. Rivera-González, 776 F.3d 45, 48, 49-51(1st Cir. 2015); United States v. Bermúdez-Meléndez, 827 F.3d 160, 166 (1st Cir. 2016); United States v. de Jesús, 831 F.3d 39, 41-44 (1st Cir. 2016); United States v. Figueroa-Quiñones, 657 F. App'x 9, 11-13 (1st Cir. 2016); United States v. Santa-Otero, 843 F.3d 547, 550, 551-52 (1st Cir. 2016); United States v. Fuentes-Echevarria, 856 F.3d 22, 24, 26 (1st Cir. 2017); United States v. Lugo- Cartagena, 701 F. App'x 6, 7, 11-12 (1st Cir. 2017); United States v. Vázquez, 854 F.3d 126, 129-30 (1st Cir. 2017); United States v. Garay-Sierra, 885 F.3d 7, 11, 15-16 (1st Cir. 2018); United States v. Hernández-Ramos, 906 F.3d 213, 214-15 (1st Cir. 2018); United States v. Laureano-Pérez, 892 F.3d 50, 52-53 (1st Cir. 2018); United States v. Severino-Pacheco, 911 F.3d 14, 17, 22 (1st Cir. 2018); United States v. Miranda-Díaz, 942 F.3d 33, 37, 42-43 (1st Cir. 2019); United States v. Viloria-Sepulveda, 921 F.3d 5, 6, 10- 11 (1st Cir. 2019); United States v. García-Mojica, 955 F.3d 187, 190-91, 193 (1st Cir. 2020); United States v. Díaz-Rivera, 957 F.3d 20, 22, 29-30 (1st Cir. 2020); United States v. Gonzalez, 988 F.3d 100, 101-03 (1st Cir. 2021); United States v. Merced-García, 24 F.4th 76, 79, 81 (1st Cir. 2022).

54

upward variance from the guidelines range of 41 to 51 months (3.42 to 4.25 years). See 955 F.3d at 192. The district judge "cited" Puerto Rico's "problem of illegal weapons." See id. at 193. But in affirming, García-Mojica noted that the judge also emphasized the defendant's "pattern of serious weapons offenses in his particular community." See id. And the judge's "articulation of these concerns," García-Mojica concluded, justified "additional deterrence" and more prison time. See id.

The bottom line is that we have long allowed judges to use community characteristics in sentencing. But even Flores- Machicote warned them against "go[ing] too far" (a warning retold many times since). Mindful of that message, all the just-listed cases held that each judge there did not go "too far" in using community characteristics to inform the upward-variance decision.

And as the government also notes, not every precedent addressing community characteristics is of the affirming sort. Another case line of ours has vacated upwardly variant sentences premised on the kind of community characteristics considered permissible in the other case line. See Carrasquillo-Sánchez, 9 F.4th at 60-62; United States v. García-Pérez, 9 F.4th 48, 52-55 (1st Cir. 2021); Rivera-Berríos, 968 F.3d at 136-37; United States v. Ortiz-Rodríguez, 789 F.3d 15, 18-19 (1st Cir. 2015). We found error there because the judges plainly based their rulings on community-centered concerns rather than on individualized

55

assessments of each defendant's circumstances. See Carrasquillo- Sánchez, 9 F.4th at 60-62; García-Pérez, 9 F.4th at 52-55; Rivera- Berríos, 968 F.3d at 136-37; Ortiz-Rodríguez, 789 F.3d at 18-19.

Rivera-Berríos — a simple gun-possession case, "a non-

violent and victimless crime," see 968 F.3d at 135 — is a leading exemplar. In picking an above-guidelines sentence — which (to remind) requires a heightened explanation — the Rivera-Berríos judge apparently "relied on nothing beyond the mere fact that the offense of conviction involved a machine gun." See id. That, in other words, was the "sole factor upon which [he] relied as the basis for the upward variance." Id. at 136 (emphasis added). But the "guideline[s]" already figured that factor into the sentencing "calculus," Rivera-Berríos said. Id. "And," Rivera-Berríos noted, "the record" lacked "any basis for giving that factor extra weight." Id.

Discussing "deterrence and punishment," the Rivera-

Berríos judge did say what he thought Puerto Rico's crime trends were. See id. at 136-37. But "[u]nmoored from any individual characteristics of either the offender or the offense of conviction," the judge's community-centric concerns could not "serve as building blocks for an upward variance." See id. at 137. And Rivera-Berríos added that — even with his many "institutional" advantages (including being perfectly positioned to consider the particularities of each individual case), see

56

Kimbrough, 552 U.S. at 109 — the judge did not identify a "special characteristic attributable either to the offender or to the offense" (beyond that it occurred in Puerto Rico) that "remove[d]" the "case from the mine-run" (and neither did the record), see Rivera-Berríos, 968 F.3d at 137. So Rivera-Berríos vacated and remanded for resentencing.

Our colleagues' opinion accuses us of plucking "out of thin air" the idea that deterrence "cannot serve by itself ('solely') to support any upward variance." But that idea comes straight from the variance-vacating decisions like Rivera-Berríos, as just shown, see also Carrasquillo-Sánchez, 9 F.4th at 60-61 — decisions that faithfully applied existing law, see Rivera- Berríos, 968 F.3d at 137 (applying (among other cases) Flores- Machicote, 706 F.3d at 21). And these variance-vacating opinions remain very much alive and controlling, having survived this en banc (because of the 3-3 split) fully intact. This idea — we cannot stress enough — also comes straight from the individualized- sentencing requirement that the Sentencing Reform Act itself imposes.

57

B

Helpfully, the parties share some common ground — not only about the judge's sentencing rationale but also about how the sentence relates to our Circuit's precedent as it currently stands.

Regarding the judge's sentencing rationale, the parties agree that the judge relied exclusively on community characteristics in varying upward from the guidelines. We say this because — using Rivera-Berríos lingo — they agreed during en banc oral argument that the judge failed to draw a "case-specific nexus" between the community characteristics and Flores's situation, beyond (of course) his machine-gun possession. And we agree with them about this — given the judge's explicit statements that • Flores's offense was not "more harmful than other[] similar"

offenses;

• the mass-shooting audio and video had nothing to do with Flores's "conduct in this case"; and • the variance came about because Flores's crime fit "within a category of offenses, gun crimes, that the [c]ourt, considering the particular situation in Puerto Rico, views as

58

more serious here than if they had occurred in a less violent society."17 Taken at face value, the judge's words and actions suggest that he would vary upward as much as he did here (and perhaps more) for every gun-crime offender in Puerto Rico (at least absent mitigating circumstances not present here), simply because he thinks Puerto Rico is awash in gun violence — without ever tying the variance's expected effects to the specifics of each offender (his comments arose in discussing the § 3553(a) factors, not in discussing any Kimbrough-style policy disagreement; remember how he left the "Policy Disagreement with the Guidelines (Kimbrough v. U.S., 552 U.S. 85 (2007)" box blank on the written statement of reasons).

Regarding our precedent, the parties agree that Rivera-

Berríos and its successors counter the judge's view that community

17 To spend a few more minutes on the judge's not-more-harmful-

than-other-similar-crimes comment: The record does not undercut it. Indeed the facts actually are "entirely consistent with simple possession of a machine gun." See Rivera-Berríos, 968 F.3d at 133, 135 (ruling that a Glock with 18 rounds and a detached magazine was not inconsistent with simple possession); García-Pérez, 9 F.4th at 54 (similar, involving a Glock with 65 rounds and 3 magazines). And no contrary inference can be properly drawn. "[W]hile 'a court's reasoning can often be inferred by comparing what was argued by the parties or contained in the presentence report with what the judge did,' such inferences must be anchored in 'what the judge did.'" Carrasquillo-Sánchez, 9 F.4th at 62 (quoting Jiménez–Beltre, 440 F.3d at 519); see also García- Pérez, 9 F.4th at 55. And what the judge did here was make explicit that he was not relying on the fact that "Flores'[s] crime itself was more harmful than others similar to his."

59

factors alone justified the upward variance sentence here — requiring us to vacate if the Rivera-Berríos family of cases stays good law. And we agree with them on that point too.

But that is where the consensus ends. Convinced that Rivera-Berríos and its like are out of step with controlling law, the government asks us to affirm Flores's sentence — despite the judge's exclusive reliance on community characteristics to support the upward variance. Said differently, the government wants us to hold that Rivera-Berríos and cases following it wrongly vacated upward variances arising from their exclusive reliance on community characteristics.

C

We start with the government's claim — apparently seconded by our colleagues' opinion — that we must affirm Flores's prison term because judges enjoy "broad" sentencing "discretion."

A huge flaw with that theory is that discretion is not code for anything goes, see Igartúa-De La Rosa v. United States, 417 F.3d 145, 149 (1st Cir. 2005) (en banc) — and so is not of itself a reason to affirm, as our review of the history of federal sentencing reveals. "Discretionary decision-making does not mean standardless decision-making," to state the obvious. See Kearney v. Standard Ins. Co., 175 F.3d 1084, 1105 (9th Cir. 1999) (Fernandez, J., dissenting) (citing Cooter & Gell v. Hartmarx Corp., 496 U.S. 384, 405 (1990)); see also Nken v. Holder, 556

60

U.S. 418, 434 (2009). A familiar example is a high school hockey "referee[], whose game calls are unappealable" but who "do[es] not act in a standardless world." See Kearney, 175 F.3d at 1105 (Fernandez, J., dissenting).

Even in the pre-guidelines days — when sentencers had far more discretion than now, with "similar offenders" doing "similar offenses" getting "different sentences" from "the same judge on different days, different judges on different days, different judges on the same day, and different judges in different jurisdictions," see Richard Singer, In Favor of "Presumptive Sentences" Set by a Sentencing Commission, 24 Crime & Delinq. 401, 402 (1978) — we still saw a need to pump the brakes. Consistent with the age-old sentencing tradition that considers each offender as an individual, Wardlaw could not have been more emphatic (as we said before, so the following quotes should be familiar). While "general deterrence" is a "legitimate" sentencing goal, Wardlaw held that judges abuse their discretion if they "'mechanistic[ally]'" conclude that certain classes of crime "invariably deserve[]" certain kinds of punishment. See 576 F.2d at 938 (quoting Foss, 501 F.2d at 527). Sentencers "holding such fixed ideas [are] presumably closed to individual mitigating factors." See id. (quoting Foss, 521 F.2d at 527). So they must "always" consider deterrence together with the defendant's "individual circumstances," avoiding the use of "questionable

61

assumption[s]," and resisting any temptation to "view[]" sentences "chiefly as instruments of retaliation against other, different [defendants]." See id. at 938-39. Tellingly, neither the government nor our colleagues' opinion cites — let alone distinguishes — Wardlaw.

Echoes of Wardlaw's caution still linger in our post-

guidelines variance opinions, whether the opinions are affirming or vacating variances. And while "[t]he allowable band of [guidelines] variance is greater" after Booker,

intellectual discipline remains vital.

"[A] motion to [a court's] discretion is a motion, not to its inclination, but to its judgment; and its judgment is to be guided by sound legal principles."

See Kirkpatrick, 589 F.3d at 416 (quoting United States v.

Burr, 25 F. Cas. 30, 35 (C.C.D. Va. 1807) (Marshall, C.J.)) (emphases added and remaining alterations by Kirkpatrick).

Our point is that neither the government nor our colleagues' opinion can simply invoke "broad" sentencing "discretion" (or the like) and declare victory. The issue to be resolved is whether judges can exercise discretion in a way that gives as much and as exclusive weight to community characteristics as Flores's judge gave them in varying upward.

D

Citing Politano, Flores-Machicote, Viloria-Sepulveda, Zapata-Vázquez, de Jesús, and Pantojas-Cruz, the government

62

separately hints that some First Circuit cases already hold that sentencers can rely solely on community concerns to vary upward — without having to root sentences in some characteristic traceable to the criminal or the crime. This is the same flavor of argument that our colleagues' opinion seemingly makes. Anyway, the government continues that stray decisions like Rivera-Berríos botched matters by slighting them. So it seems the government wants us to affirm those earlier rulings and make clear that Rivera-Berríos and cases of its type wrongly deviated from binding First Circuit precedent permitting such exercises of sentencing discretion.

This argument is unconvincing.

The cases hyped by the government and our colleagues'

opinion — while recognizing that sentencers may "take into account the characteristics of the community in which the crime took place" — themselves hold that judges must not "go too far" (or similar words). See Zapata-Vázquez, 778 F.3d at 23-24 (emphasis added and quotation marks omitted); see also Viloria-Sepulveda, 921 F.3d at 10-11; de Jesús, 831 F.3d at 41-44; Pantojas-Cruz, 800 F.3d at 59- 60; Flores-Machicote, 706 F.3d at 23-24; Politano, 522 F.3d at 73- 74. And also much like Rivera-Berríos and its heirs, the government's and our colleagues' opinion's preferred cases also hold that judges go "too far" when they "focus too much on the community and too little on the individual" (or similar language).

63

See Flores-Machicote, 706 F.3d at 24; see also Viloria-Sepulveda, 921 F.3d at 10-11; de Jesús, 831 F.3d at 41-44; Pantojas-Cruz, 800 F.3d at 59-60; Politano, 522 F.3d at 73-74. Importantly too, none of these cases relied solely on community characteristics in varying upward. See Flores-Machicote, 706 F.3d at 24 (explaining that the judge also relied on a finding of "likely recidivism"); Viloria-Sepulveda, 921 F.3d at 9-11 (explicating that the judge relied on a finding based on images on the defendant's phone that the defendant was "associat[ed] with violent and illegal conduct"); de Jesús, 831 F.3d at 43 (clarifying that the judge "use[d] the Puerto Rican crime rate as one of several integers"); Pantojas-Cruz, 800 F.3d at 60 (saying that the judge also relied on a finding that a local court had found "probable cause against [the defendant] for a murder committed with the weapon he was [federally] charged with possessing"); Politano, 522 F.3d at 74 (noting that the judge relied on the fact that the defendant had "more encounters with . . . law enforcement than [were] countable" and was likely to "recidiv[ate]").

So we see no reason to hold that our pre-Rivera-Berríos decisions already let judges vary upward in gun-possession cases based solely on a concern about the crime rate in a community as big as Puerto Rico when that concern is "unmoored" from any offender or offense-specific characteristic (as was the case in Rivera-Berríos and is the case here).

64

E

Additionally but relatedly, the government claims that Rivera-Berríos and like opinions "conflict with" Flores-Machicote itself and so should not be relied on. This is a centerpiece feature of the government's efforts here (above we previewed how the government predicated its en banc petition on this "conflict" notion).

The suggestion is off base, largely for reasons we have already explained.

Both groups of cases — variance-affirming cases like Flores-Machicote and variance-vacating cases like Rivera-Berríos — hold that judges can consider community-based concerns in sentencing. See Flores-Machicote, 706 F.3d at 22-23; Rivera- Berríos, 968 F.3d at 136; Carrasquillo-Sánchez, 9 F.4th at 59-60. And both groups also hold that judges can abuse their discretion by focusing "too much on the community and too little on the individual." See Flores-Machicote, 706 F.3d at 24; Rivera- Berríos, 968 F.3d at 136-37; Carrasquillo-Sánchez, 9 F.4th at 59- 63.18

Our colleagues' opinion also accuses us of pulling the "too 18

much" restriction out of the sky. But the reader can see that that limitation comes from our caselaw. Pointing to the conclusion at our opinion's end — that we would vacate Flores's sentence and remand for resentencing within the advisory prison range of 24 to 30 months, effectively barring any variance here, see section V — our colleagues' opinion also thinks that we think that any variance "greater than zero" is "too much" in any case. But that is simply

65

The question then is whether a sentence like Flores's focused too much on the community and too little on the individual. But regrettably for the government, Flores-Machicote does not answer that question because it had to focus on a different one — whether judges can consider the perceived ineffectiveness of local courts and local violent-crime rates to "inform[] and contextualize[] the relevant need" for general deterrence in varying upward (which again the judge did after also emphasizing the defendant's significant criminal history and finding a likelihood of recidivism). See 706 F.3d at 23-24.

F

Still thinking that Flores's sentence would prevail under our pre-Rivera-Berríos cases, the government lists decisions saying that sentencers need not give every § 3553(a) factor equal billing — thus allowing them to emphasize community concerns (like general deterrence) over individual ones (like a defendant's background). And from there the government protests that we must rein in cases like Rivera-Berríos (what with their supposedly wayward approach) and affirm Flores's sentence, because there is no way to square Rivera-Berríos's analysis with cases of ours

incorrect. We do not question any of our variance-affirming precedents. And we would have a within-range sentence in this matter because the government proposed one below, plus the judge thought Flores's case was not more harmful than others like his.

66

saying that general-deterrence concerns may support an upward variance.

Color us unpersuaded.

The very decisions the government cites — after saying that sentencers "need not afford equal weight to each [§ 3553(a)] factor in a given case" — hold that judges "go too far" (there is that phrase again) if they overemphasize "community-based" factors at the cost of "case-specific" ones (or words to the same effect). See Zapata-Vázquez, 778 F.3d at 24 (emphasis added and quoting Flores-Machicote, 706 F.3d at 23-24); see also Viloria-Sepulveda, 921 F.3d at 10-11; de Jesús, 831 F.3d at 41-44; Pantojas-Cruz, 800 F.3d at 59-60; Politano, 522 F.3d at 73-74. And again none of those decisions sanctioned upward variances centered solely on a finding that a community characteristic called for extra deterrence — as none addressed upward variances based on such a singular ground.

G

1

Having rejected the notion that our pre-Rivera-Berríos cases require that we affirm Flores's sentence, we now turn to the government's claim that the panel here and the panels in the Rivera-Berríos class of cases gaffed things because the challenged judges actually "individualized" the sentences and so acted in a procedurally reasonable way. Our colleagues' opinion also claims

67

that Flores's judge used an "individualized" approach. As support, the government and our colleagues' opinion note the judge mentioned the § 3553(a) considerations (including some mitigating facts) but decided the need to deter violent gun crime in Puerto Rico generally (itself a § 3553(a) concern) was so important that it alone called for a substantially higher sentence than the guidelines suggested. And the government and our colleagues' opinion also say or imply that the judge focused on generally deterring gun violence only in the community where Flores chose to commit the crime, thus making the sentence individualized.

Supreme Court decisions cut against their positions.

As the high Court has been at pains to stress, particularized facts about the offender matter greatly in sentencing under the Sentencing Reform Act. See, e.g., Gall, 552 U.S. at 54 (stating that "the unique facts" of the defendant's situation supported the judge's decision to give a below- guidelines sentence). See generally Concepción v. United States, 142 S. Ct. 2389, 2395 (2022) (repeating that "a judge at sentencing considers the whole person before him or her 'as an individual'" (quoting Koon, 518 at 113)). And when it comes to the § 3553(a) factor-weighing, the Court has also been at pains to note that sentencers are "in a superior position to find facts and judge their import" in deciding the most appropriate sentence for a given defendant, because they "see[] and hear[] the evidence, make[]

68

credibility determinations, ha[ve] full knowledge of the facts and gain[] insights not conveyed by the record." See Gall, 552 U.S. at 51 (quotation marks omitted). Which gives them "access to, and greater familiarity with, the individual case and the individual defendant before [each of them] than the [c]ommission or the appeals court," see id. at 51-52 (emphases added and quoting Rita, 551 U.S. at 357-58) — something that ensures that "the punishment . . . suit[s] not merely the offense but the individual defendant," see Wasman v. United States, 468 U.S. 559, 564 (1984) (emphasis added).

Given these differing roles, the Court has stressed that a judge must always conduct an "individualized assessment" based on the facts presented. Gall, 552 U.S. at 50. And the Court has also stressed that varying from the commission-selected range risks creating a disparity between the defendant and others with similar records and offenses, because that range is the one the commission (exercising its own institutional strengths) decided "might achieve § 3553(a)'s objectives" in mine-run cases. See Rita, 551 U.S. at 350-51; accord Kirkpatrick, 589 F.3d at 415 (noting that "[w]henever a court gives a sentence substantially different from the [g]uidelines' range, it risks creating unwarranted sentencing disparities . . . for most other judges will give sentences closer to the norm," and adding that "[t]hat's a major reason why substantial variances from the . . .

69

[c]ommission's recommendations require careful thought" (citing Gall, 552 U.S. 38; Spears, 129 S. Ct. 840; Nelson, 129 S. Ct. 890)). Which is why the Sentencing Reform Act requires that when judges sentence outside the guidelines range, they must provide a "justification" based on their "individualized assessment" that is "sufficiently compelling to support the degree of the variance." Gall, 552 U.S. at 50; see also Nelson, 555 U.S. at 351.

Flashing back again to the pre-guidelines years, we know that our court and others said that "[i]n each case, a criminal sentence must reflect an individualized assessment of a particular defendant's culpability rather than a mechanistic application of a given sentence to a given category of crime." See United States v. Barker, 771 F.2d 1362, 1365 (9th Cir. 1985) (citing Williams, 337 U.S. at 247; Foss, 501 F.2d at 527; and Wardlaw, 576 F.2d at 938). So (for instance) our Wardlaw opinion held that giving "mules" harsher sentences based on an "unbending" and "questionable" notion that it would force "mule owners" to do "dirty work" that would lead to their arrest was not reasonable because it was not "individualized." See 576 F.2d at 936, 937, 938 (citing Foss, 501 F.2d at 527). And similarly, the Ninth Circuit's Barker opinion held that "the desire to stem the tide of marijuana smuggling through the deterrent effect" on other would- be smugglers could not alone serve as the judge's reason for imposing a five-year sentence on drug smugglers (instead of the

70

government's recommended 18-month sentence). See 771 F.2d at 1367- 69.

Barker rightly said that the "desire to 'send a message'

through sentencing [is not] inappropriate per se." See id. at 1368. But Barker also rightly held that message-sending is "subject to limitation" in that judges must always "balance[]" it in light of the defendant's individual circumstances. See id. at 1369. That is because

[c]entral to our system of values and implicit in the requirement of individualized sentencing is the categorical imperative that no person may be used merely as an instrument of social policy, that human beings are to be treated not simply as means to a social end like deterrence, but also — and always — as ends in themselves.

Id. at 1368-69; accord Concepción, 142 S. Ct. at 2399.

Quoting us out of context, our colleagues' opinion says that we believe "general deterrence [is] a 'questionable' rationale." Hardly. We know as well as anyone that general deterrence is a relevant sentencing factor. And our "questionable" quote actually comes from Wardlaw, where we doubted that the district court's specific mules-based punishment theory would achieve general deterrence.

True to first principles, we have never affirmed an upward variance like Flores's — one for a firearms-possession crime and powered just by a locale's high violent-crime rate, with no

71

judge-made finding tying the offender and his offense to that community characteristic through his means of committing the offense. And when we asked the government's lawyer at oral argument whether he knew of any gun-possession case outside this Circuit where the court had, counsel said he knew of no such case. Our research has not turned up one either.

Against this background, we cannot accept the claim that the judge based Flores's variant sentence on an individualized assessment or used "community characteristics" only to "contextualize and inform" the "relevant" need for general deterrence in Flores's individual case. See Flores-Machicote, 706 F.3d at 23-24. No one can doubt (at least no one should doubt) that the sentence rests exclusively on a general-deterrence rationale that — because the judge tied it to Puerto Rico's violent-crime rate — would be decisive in any machine-gun- possession case prosecuted there (absent mitigating circumstances not present here). Yet as we said, the judge flatly stated that Flores's possession was no more "harmful than others similar to his." And the judge explicitly declined finding that Flores's "conduct in this case" was associated with the community-based gun-violence concern that alone drove the upward variance. So the judge made no finding that Flores is more associated with or more prone to commit the kind of violence in the community requiring

72

deterrence than others possessing a gun anywhere in the country.19 And as a result we conclude that the judge — based on his own explanation — varied upward by "focus[ing] too much on the community and too little on the individual." See id.

As our colleagues' opinion notes, the record shows that officers found ammo and a spent casing in the car Flores was in at the time of his arrest. If the judge had made not-clearly- erroneous findings that Flores likely used the gun or otherwise added to the violence in Puerto Rico, the question before us might be different — as likely would the total offense level and thus the guidelines range in his case, see USSG § 2K2.1(b). Our analysis might also change if the judge had factored in how (as our colleagues' opinion notes) Flores got arrested with a machine gun at a McDonald's parking lot. But the inescapable truth is that the judge made no such findings (or anything similar) and instead found Flores's offense no more harmful than similar offenses, a finding neither the government nor our colleagues' opinion says is clearly erroneous.

That the judge also recognized other § 3553(a) factors favoring Flores does not alter our conclusion. Neither we nor the

19 For that reason we reject the government's claim that Flores's case — where the geographic area covers all of Puerto Rico — is similar to one where a judge might find a machine-gunpossessing defendant culpable because he possessed it on or near school grounds or another sensitive setting.

73

Supreme Court has ever held that a judge satisfies the procedural duty to explain why a sentence fits the particular offender simply by noting mitigating factors exist in a particular case. See Rita, 551 U.S. at 356-58. That is especially so where "the judge imposes a sentence outside" the range identified for the mine-run case. See id.; see also Rivera-Berríos, 968 F.3d at 136–37 (noting that "the mere fact that the court considered all of the relevant factors cannot justify an upward variance when those factors, whether taken singly or in combination, do not form a permissible basis for an upward variance" (citing Flores-Machicote, 706 F.3d at 21)); accord Chavez-Meza v. United States, 138 S. Ct. 1959, 1965 (2018) (citing Rita, 551 U.S. at 357).20 Do not get us wrong. We are not saying judges would be focusing too much on the community and too little on the individual if they deemed a defendant more culpable than the typical offender because he did the crime with some intent to go to a particular area and act in a way that exacerbated the conduct requiring deterrence. A variance in that scenario would be individualized precisely because it would be "grounded" in a finding reflecting the individual's heightened culpability in his particular case.

20 While we give "some weight" to a judge's statement about having "considered" the § 3553(a) factors, Dávila-González, 595 F.3d at 49 (citing United States v. Morales–Machuca, 546 F.3d 13, 26 (1st Cir. 2008)), we still must decide whether the judge applied them in a reasonable way.

74

And that is so because it would represent a finding that the defendant was "associat[ed] with the violent and illegal conduct" plaguing the community. See Viloria-Sepulveda, 921 F.3d at 9 (citing United States v. Acevedo-Lopez, 873 F.3d 330, 340 (1st Cir. 2017); United States v. Quiñones-Meléndez, 791 F.3d 201, 205 (1st Cir. 2015); and United States v. Gallardo-Ortiz, 666 F.3d 808, 815 (1st Cir. 2012)).

Consider Viloria-Sepulveda. In that machine-gun-

possession case, we affirmed an upward variance because the judge committed no clear error in finding that photos on a defendant's cell phone — showing him and others carrying guns — "signaled his past participation in or propensity for illegal or violent activities involving drugs and firearms" in Puerto Rico. Id. And we notably said as well that the judge's finding helped "'contextualize[]'" his expressed concern about the "pervasiveness of guns and the level of violence" there, such that we could rule that he did not "overemphasize these community concerns at the expense of individual ones." See id. at 10 (quoting Flores- Machicote, 706 F.3d at 23). Compare the situation in Viloria- Sepulveda to the situation here and the difference is night and day.

We must add that this approach — basing a variance on an individualized assessment of whether the defendant culpably added to the conduct in the community needing deterrence — may help solve

75

the puzzle of what exactly is a "community" for community- characteristics purposes: a state or a commonwealth like Puerto Rico, a county, a city, a town, etc.? We say this because in practice the relevant community should become clear if what counts is the defendant's relationship — individual to him — to that community. We actually asked the parties how to define "community" in this context. See Flores-González, 46 F.4th at 60 (mem.). And they apparently agreed that it depends on the particular circumstances of the particular case.21 But because this is a fact-bound issue, we leave it to be hashed out in the district courts (if necessary).

2

The government also insists that a rule stopping judges from varying upwards based solely on community-centered concerns not tied to case-specific factors puts us in "substantial tension" with two out-of-circuit opinions: United States v. Hatch, 909

Flores says that the definition "depend[s] on the district 21

court's reasoning regarding the characteristic's significance and its relationship to a valid sentencing factor." The government says that "depending on the circumstances of the particular case, a sentencing court could reasonably determine that the relevant community is a single state or the Commonwealth of Puerto Rico, a county, a city, a town, or something else, such as a discernable region within a state or a region covering multiple states."

76

F.3d 872 (7th Cir. 2018) (per curiam), and United States v. Cavera, 550 F.3d 180 (2d Cir. 2008) (en banc).

We disagree.

The Seventh Circuit's Hatch opinion affirmed an upward variance for a gun-trafficking defendant. See 909 F.3d at 874. The judge there had said "that the rise of gun violence in Chicago meant that the [s]entencing [g]uidelines did not adequately reflect the seriousness of [the defendant's] offense or sufficiently deter firearm trafficking." Id. But the judge had also found that • the defendant had "illegally brought handguns into Chicago three times," "[t]he first time[] accompanied by . . . a large-scale Chicago gun dealer";

• he had used a friend to buy guns in another state to do so;

and • "altogether" he had trafficked 17 guns into Chicago's black market, some going to "felons" and even "a minor."

Id. at 874-75. Seeing no reversible error, the Seventh Circuit wrote:

Beyond geographic issues, the judge considered [the defendant's] history and characteristics (mostly "in his favor"), the nature of the offense ("troubling," and [the defendant's] failure to fully accept responsibility "bothered" him), the seriousness of the offense ("difficult to

77

overstate"), and the need for deterrence and respect for the law.

Id. at 875.

The Second Circuit's en banc Cavera opinion also affirmed an upward variance for a gun-trafficking defendant. See 550 F.3d at 184, 197. Seeking "to accomplish the goal of general deterrence," the district judge — focusing on the case-specific context, so as to make "an individualized judgment" — had noted that • New York City had a "profitable black market in firearms,"

created by the state's "strict gun control laws";

• the defendant had sold 16 guns to New York dealers just before his arrest; and • "the consequences for the community of bringing or transporting . . . firearms into New York City" included the increased risk of future violence.

See id. at 185, 186, 197. Wrapping up, the Second Circuit said that the defendant "knew the guns he sold were destined for New York." Id. at 197 (emphasis added). So "he was a knowing participant in the traffic heading in that direction" — meaning that "[a]s a result, there was no abuse of discretion in the [judge's] decision to consider New York market conditions . . . to

78

accomplish the goal of general deterrence." Id. (emphases added).22 What stands out in bold relief is that each of these sentencers — unlike Flores's — "ground[]" an upward variance "in case-specific factors," see Flores-Machicote, 706 F.3d at 24, with the record showing (to focus again on Cavera) that the defendant "knowing[ly]" trafficked the guns into the relevant community and thus helped add to the gun violence there, see Cavera 550 F.2d at 197 (emphasis added); see also Hatch, 909 F.3d at 874.

H

The government makes three more far-reaching claims for why we must affirm Flores's sentence. Our colleagues' opinion joins in one of them.

None succeeds.

1

The first of these arguments focuses on our Circuit's requirement (mentioned earlier in discussing Rivera-Berríos) that if "a sentencing court relies on a factor already accounted for by

22One should know that Judge (now Justice) Sotomayor — in an opinion concurring and dissenting in part — said that if it is true "that the black market for guns is only profitable" in a few urban areas like New York City, then it is also true that the guidelines "already account for any deterrence issues raised by New York's strict gun laws," because gun-trafficking crimes happen "almost entirely or predominantly in those areas." See id. at 222 (Sotomayor, J., joined by Cardamone and Straub, JJ., concurring and dissenting in part, and by Pooler, J., in part).

79

the sentencing guidelines to impose a variant sentence, [it] must indicate what makes that factor worthy of extra weight." See United States v. Díaz-Lugo, 963 F.3d 145, 155 (1st Cir. 2020) (alteration in original and quotation marks omitted); see also Zapete-García, 447 F.3d at 60. To hear the government tell it, this requirement clashes with Gall and its siblings because (to quote the government) the rule "effectively operates as an erroneous presumption that a variance from the [g]uidelines is categorically unreasonable."

Not so.

Every time judges pick "sentence[s] substantially different from the [g]uidelines' range, [they] risk[] creating unwarranted sentencing disparities, in violation of 18 U.S.C. § 3553(a)(6), for most other judges will give sentences closer to the norm." Kirkpatrick, 589 F.3d at 415.23 So "substantial variances from the . . . [c]ommission's recommendations require careful thought." Id. (citing Gall and Spears, among other caselaw). And while the Supreme Court forbids us from requiring "'extraordinary' circumstances to justify" an outside-guidelines sentence, we know that a major variance demands a more substantial

23 The government also claims that disparity concerns vanish when a defendant possesses a gun in a community the judge concludes is "atypically plagued by that offense." But as we will see (in section IV.H.3(c)), the government offers no convincing rationale or standard for such a judge-defined plagued-by-crime rule.

80

reason than a minor one — in Gall's phrasing, the reason must be "sufficiently compelling to support the degree of the variance" (it is fair to say then that a substantial variance ups the chances of appellate reversal). See 552 U.S. at 47 (first quote), 50 (second quote) (emphases added); see also id. at 47-51 (discussing meaningful appellate review for reasonableness). See generally Rita, 551 U.S. at 354, 356 (noting that a sentencer must "set forth enough to satisfy the appellate court that he has . . . a reasoned basis for exercising his own legal decisionmaking authority" and noting that "Circuit courts exist to correct . . . mistakes when they occur"). The extra-weight requirement is just the sufficiently-compelling-reason rule in action.

A post-Gall case of ours — United States v. Ofray-Campos, 534 F.3d 1 (1st Cir. 2008) — nicely illustrates the point. Convicted of conspiring to distribute drugs, a defendant there received a sentence "two and one half times greater — and more than twenty-four years longer — than the top of the recommended guidelines range." Id. at 42-43. Two factors propelled the variance: "(1) [the defendant's] possession of 'powerful weapons' as a 'triggerman,' and (2) his involvement in violence in connection with the narcotics activity." Id. at 43. Those factors, we said, "may have justified a substantial upward variance" — just not one as big as he got. Id. And we took particular issue with the judge's "triggerman" comment, noting

81

that the "firearm possession had already been considered, and accounted for," in a guidelines enhancement that raised the defendant's offense level. See id. Repeating Gall's warning that a sentencer's explanation must match the variance's degree, we then said that this factor was "not so distinct from the firearm possession that was incorporated into the guidelines calculation as to justify a variance of such magnitude." Id. at 43 (emphases added and citing Zapete-García, 447 F.3d at 60). See generally United States v. Flores-Nater, 62 F.4th 652, 657 (1st Cir. 2023) (stating that an "upward variance must rest on more than factors already accounted for in the guideline[s] calculus").

Casting its gaze elsewhere, the government also claims that our extra-weight requirement is out of sync with Sixth, Tenth, Eleventh, and D.C. Circuit caselaw. But decisions from those Circuits suggest otherwise. To quote a typical case — which also happens to quote our Zapete-García opinion — the Eleventh Circuit has held that judges "may rely on a factor 'already included in the calculation of the guidelines sentencing range' so long as" they state "'specifically the reasons that [their] particular defendant's situation is different from the ordinary situation covered by the guidelines calculation.'" See United States v. Styles, No. 20-13321, 2021 WL 4059953, at *4 (11th Cir. Sept. 7, 2021) (per curiam) (emphasis added and quoting Zapete-García, 447 F.3d at 60); see also United States v. Boucher, 937 F.3d 702, 708-

82

09 (6th Cir. 2019) (noting that the court has vacated an above- guidelines sentence based on a factor "already incorporated into the [g]uidelines-recommended sentence" because the judge did not sufficiently explain what distinguished the defendant's case from the usual one); United States v. Alapizco-Valenzuela, 546 F.3d 1208, 1222-23 (10th Cir. 2008) (stating that if judges impose variant sentences based on factors "already accounted for in the advisory [g]uidelines range," they must state "specifically the reasons that [their] particular defendant's situation is different from the ordinary situation covered by the guidelines calculation" (emphasis added and quoting a case that quotes Zapete-García, 447 F.3d at 60)); United States v. Brown, 808 F.3d 865, 873 (D.C. Cir. 2015) (declaring that judges can vary based on factors already accounted for by the guidelines if they show "how . . . the [g]uidelines do not fully account for those factors" (quotation marks omitted)). And none of those cases involved a judge using community characteristics to vary upward, like what happened here.

2

Thinking creatively, the government next tries to rebrand variances driven solely by community characteristics as Kimbrough variances. Reduced to basics, the government's Kimbrough-based argument runs (at least implicitly) this way. (1) Judges varying from a commission-suggested range based solely on community concerns are disagreeing with the range itself.

83

(2) Kimbrough says that judges can disagree with the commission (but not with a statute, naturally, and they must act reasonably). (3) So Kimbrough means — contrary to Rivera-Berríos and its ilk — that judges can vary based solely on community characteristics of the crime's locale (subject to the respecting-statutes and acting- reasonably caveats just mentioned).

The thesis does not hold together (to be fair a case of ours dropped a footnote suggesting the possibility that a variant sentence driven only by this kind of community characteristic might be a Kimbrough variance, see Carrasquillo-Sánchez, 9 F.4th at 61 n.2 — but the answer to that suggestion is no, for reasons we are about to come to).

Unhappy with rampant sentencing disparities under the old regime, Congress tasked the commission with creating "sentencing policies and practices for the Federal criminal justice system." See 28 U.S.C. § 991(b)(1) (emphases added). Staffed by experts, the commission knows that any actual crime will be done only in a particular place, at a particular time, in a particular way, and by a particular offender with a particular background. Cf. United States v. Aguilar-Peña, 887 F.2d 347, 351 (1st Cir. 1989) (noting that "[b]ecause the grounds for departure derived their essence from the offense itself, not from [idiosyncratic] circumstances attendant to a particular defendant's commission of a particular crime, the grounds,

84

virtually by definition, fell within the heartland"). But a commission-endorsed guidelines range is generally meant to apply in any case involving that offense. See Rita, 551 U.S. at 350. That is why the commission expects judges "to treat each guideline as carving out a 'heartland,' a set of typical cases embodying the conduct that each guideline describes." See USSG Ch. 1 Pt. A, introductory cmt. 4(b) (emphasis added). So the range reflects a judgment about the right range for the mine-run way of committing the crime, not a judgment about the right range for every case involving that crime.

All of which is to say that an outside-guidelines sentence powered solely by community concerns does not reflect a policy-based beef with the commission's reasons for setting the range. It simply reflects a decision that the case is not mine- run and so is not one for which that baseline range was established. Otherwise — if it were as the government suggests — then one could call any variant sentence an exercise of Kimbrough authority, which would make the sentence reverse-proof.24

24If more were needed — and we doubt that it is — Kimbrough (as we said a few pages ago) reinstated a below-guidelines sentence for crack possession after noting (among other things) the "[c]ommission's consistent and emphatic position that the crack/powder disparity is at odds with § 3553(a)." See 552 U.S. at 111. But here the government points us to nothing suggesting the commission has expressed such a concern with USSG § 2K2.1(a)(4)(b) — the provision underpinning Flores's base offense level.

85

Leaning on Concepción — a fairly recent Supreme Court case — the government (quoting that decision) finally says that because sentencers have "wide discretion in the sources and types of evidence" that they may use, see 142 S. Ct. at 2395-96 (quoting Williams, 337 U.S. at 246), they (to quote the government's brief) can "rely upon [that discretion] in determining whether a particular community possesses characteristics that would justify an upward variance."

The argument is not a difference-maker.

(a)

Concepción — which addressed sentence modifications under the First Step Act, see id. at 2396, not community-based variances — talked about judges' discretion to consider different evidence sources and types in assessing "the whole person before them," see id. at 2398 (emphasis added); see also id. at 2398-99 (noting the history of judicial discretion to consider the "fullest information possible concerning the defendant's life and characteristics" (emphasis added)). Again, our judge relied not on Flores's actions or his characteristics but on his (the judge's)

86

perception of violent crime in Puerto Rico generally to vary upward. And that distinction makes a world of difference.

(b)

Defending the judge's approach, the government and our colleagues' opinion quote back to us Concepción's comments that "[t]he only limitations on [a sentencer's] discretion" about information sources and types "are those set forth by Congress in a statute or by the Constitution." See id. at 2400. Sure. But the principle that judges may discretionarily consider a vast array of materials at sentencing does not mean that they can do so in an unreasonable way. And once again, it is unreasonable to go "too far" in relying on a community characteristic in upwardly varying a sentence. That limitation seems especially warranted when considering data about geographical differences. "Gut feelings about regional differences can be subjective in dangerous ways. Empirical data should be scrutinized because they make subjective feelings appear plausible, even when the analysis suffers from significant flaws." Cavera, 550 F.3d at 224 (Sotomayor, J., joined by Cardamone and Straub, JJ., concurring and dissenting in part, and by Pooler, J., in part); see also id. at 195 (stating that "a district court should not rely on 'subjective considerations such as "local mores" or feelings about a particular type of crime'"). See generally United States v. Colón-Maldonado, 953 F.3d 1, 10 (1st Cir. 2020) (holding that due process demands that judges not

87

sentence defendants using "materially untrue" factual "assumptions" (quoting Townsend v. Burke, 334 U.S. 736, 740-41 (1948))). Flores's judge cited no information source to corroborate his (the judge's) variance-justifying hypotheses, including (for instance) that "crime in Puerto Rico far exceeds the known limits on the mainland" and that "gun crimes" are thus "more serious here than if they had occurred in a less violent society." As the judge pointed out, the prosecutor did "mention[]" that Puerto Rico "is a hotspot for gun violence." But the prosecutor — like the judge — also offered no support for that supposed fact.

Trying to downplay our reliability concerns, our colleagues' opinion quotes a case of ours that quoted a district judge's comment calling Puerto Rico a "hot spot for weapons." See Viloria-Sepulveda, 921 F.3d at 10. But that case cites no source supporting the judge's comment and gives no indication that any party questioned the reliability of the judge's perceptions about local gun violence. Our colleagues' opinion also says that Flores "doubly waived" any reliability argument by not raising it below or in his opening brief. But we can relax a raise-or-waive rule to avoid a miscarriage of justice. See Sindi v. El-Moslimany, 896 F.3d 1, 30 (1st Cir. 2018). Regardless — and for reasons already explained — we need not resolve whether the judge violated due process by relying on the just-described findings.

88

(c)

That is a nice segue to talk about statistics. Again citing no sources, the judge mentioned "[t]he number of murders" on the island and how they have "gone down drastically from 2011." Never mind that Flores's case does not involve murder (parenthetically, no one has presented relevant statistics about machine-gun possession). The difficulty is that "[s]tatistical evidence that fails to satisfy minimum standards of reliability proves nothing." See Flores-Machicote, 706 F.3d at 24. With that in mind, we asked both sides "what . . . the current violent-crime statistics for the major municipalities in Puerto Rico" show and "[h]ow . . . they compare with other major municipalities in the United States." See Flores-González, 46 F.4th at 60 (mem.). Each said that "FBI crime data" is "the most reliable data available." But they did and do disagree about what the data means. Flores claims that "the most complete data available on violent-crime statistics" shows that "Puerto Rico as a whole" and its "major municipalities . . . have low violent-crime rates when compared with the states on the mainland and their major municipalities." Conversely the government claims that the data shows that "Puerto Rico has a particularly high rate of murder and nonnegligent manslaughter" compared to other districts. Because federal appellate courts are not factfinders, see Pullman-Standard v.

89

Swint, 456 U.S. 273, 291-92 (1982), that is a subject to be worked out in the district courts (if necessary).

Which makes this as good a place as any to say that the guidelines recognize that in some cases "a[] factor important to the sentencing determination [might be] reasonably in dispute." See USSG § 6A1.3(a). And in that scenario "the parties shall be given an adequate opportunity to present information to the court regarding that factor," see id. — with the judge required to rule on the dispute or say that the dispute does not matter, see Fed. R. Crim. P. 32(i)(3)(B), after possibly allowing "the parties to introduce evidence on the objections," see id. 32(i)(2).

I

The short of this long discussion is that none of the government's reasons why we should overrule variance-vacating opinions like Rivera-Berríos is a needle-mover (none of those decisions involved Kimbrough-policy variances, don't forget). And none of our colleagues' opinion's views (some of which mirror the government's) is a game-changer either.

V

The net effect of our en banc review is this.

As before, judges can still use geographic concerns —

like "the incidence of particular crimes in the relevant community" — to "contextualize[] the relevant need for deterrence." See Flores-Machicote, 706 F.3d at 23. But they must still "ground

90

sentencing determinations in case-specific factors" and not "focus too much on the community and too little on the individual," see id. at 24 — i.e., they cannot give so little weight to the individualized circumstances of how the defendant committed the crime that they (in the name of "general deterrence") treat every commission of it as requiring an upward variance (absent mitigating circumstances). See id.; see also Rivera-Berríos, 968 F.3d at 136-37.

And with these opinions still on the books, Flores's upward variance — lacking as it does that necessary case-specific connection — should not stand. So this court should (and we would) vacate the disputed sentence and remand for resentencing within the advisory prison range of 24 to 30 months — "within" being appropriate because the government below recommended a within- range term and the judge himself stressed that Flores's case was not more harmful than others like his. See Rivera-Berríos, 968 F.3d at 137 (taking a similar approach in a similar situation); see also United States v. Ramos-Carreras, 59 F.4th 1, 8 n.6 (1st Cir. 2023). But the grant of rehearing en banc (which vacated the prior panel's opinion) and an evenly divided en banc court (which affirms the erroneous variance by operation of law) means that Flores (unlike others) will not get the benefit of this pre- existing and still-binding precedent.