U.S. v. CAVERA

REENA RAGGI, Circuit Judge, with whom Chief Judge JACOBS, Judge CABRANES, and Judge B.D. PARKER, JR., join, concurring:

Our court, sitting en banc, today identifies a set of fundamental principles regarding the imposition and review of federal sentences after the Supreme Court's decisions in Gall v. United States, ___ U.S. ___, 128 S.Ct. 586, 169 L.Ed.2d 445 (2007), and Kimbrough v. United States, ___ U.S. ___, 128 S.Ct. 558, 169 L.Ed.2d 481 (2007). I join in the opinion of the court unanimously endorsing those principles. See ante Parts I II.A. I also join a majority of the court in holding that a 24-month sentence for illegally trafficking 16 guns to New York City is reasonable in this case and, therefore, affirming the judgment of the district court. See ante at Part II.B. I write separately simply to clarify that my rationale for affirming is not limited to the majority's approval of the district court's assessment of the need for adequate deterrence of gun trafficking into New York City. See 18 U.S.C. § 3553(a)(2)(A). Consistent with the principles announced by the court today, I conclude that the district court also acted well within its discretion in finding that local circumstances rendered Cavera's gun trafficking more serious than the mine-run case reflected in the Sentencing Guidelines' recommended 12-18 month prison range. See id. Because the majority opinion does not address this part of the district court's analysis, and because dissenting colleagues identify error therein, I here explain my reasons for concluding otherwise.

1. The Presumptions Proscribed and Review Standard Mandated by Gall and Kimbrough

In reviewing the reasonableness of the challenged sentence, two rulings in Gall v. United States, ___ U.S. ___, 128 S.Ct. 586, 169 L.Ed.2d 445, are particularly instructive. First, while a district court is statutorily obliged to consider the Guidelines in imposing sentence, see 18 U.S.C. § 3553(a)(4), it commits constitutional error if it presumes the reasonableness of a Guidelines sentence and fails to determine for itself the sentence warranted by the totality of factors stated in 18 U.S.C. § 3553. See Gall v. United States, 128 S.Ct. at 596-97; ante at 189. Second, a reviewing court is constitutionally prohibited from applying "a presumption of unreasonableness" to a non-Guidelines sentence. Id. at 597; ante at 190. These twin rulings caution that, whatever concerns reviewing courts may have about affording district courts leeway to impose non-Guidelines sentences, particularly based on general rather than case-specific concerns, we cannot, consistent with the Sixth Amendment, pronounce rules or impose burdens that effectively place a "thumb on the scales" in favor of Guidelines sentences. see Kimbrough v. United States, 128 S.Ct. at 577 (Scalia, J., concurring).

Specifically, we cannot demand "`extraordinary' circumstances to justify a sentence outside the Guidelines range." Gall v. United States, 128 S.Ct. at 595; ante at 190-91. We cannot use "the percentage of a departure [from the Guidelines] as the standard for determining the strength of the justifications required for a specific sentence." Gall v. United States, 128 S.Ct. at 595; see ante at 190-91. We cannot prohibit non-Guidelines sentences based on a sentencing judge's disagreement with Commission policy determinations. See Kimbrough v. United States, 128 S.Ct. at 570; Rita v. United States, 551 U.S. 338, 127 S.Ct. 2456, 2465, 168 L.Ed.2d 203 (2007) (observing that district courts may consider arguments that "the Guidelines sentence itself fails properly to reflect § 3553(a) considerations"); ante at 191. Finally, although we may (1) reasonably expect a "major departure [to] be supported by a more significant justification than a minor one," Gall v. United States, 128 S.Ct. at 597, and (2) conduct "closer review" of a justification "when the sentencing judge varies from the Guidelines based solely on the judge's view that the Guidelines range fails properly to reflect § 3553(a) considerations even in a mine-run case," Kimbrough v. United States, 128 S.Ct. at 575 (internal quotation marks omitted), the standard of review remains the same in every case. As Gall makes clear, "the abuse-of-discretion standard of review applies to appellate review of all sentencing decisions — whether inside or outside the Guidelines range." 128 S.Ct. at 596 (emphasis added).

2. The Procedural and Substantive Components of Reasonableness Review Gall confirms that reasonableness review has both a procedural and substantive component. At the procedural stage, Gall provides five examples of potential errors by a sentencing court: (1) "failing to calculate (or improperly calculating) the Guidelines range," (2) "treating the Guidelines as mandatory," (3) "failing to consider the § 3553(a) factors," (4) "selecting a sentence based on clearly erroneous facts," or (5) "failing to adequately explain the chosen sentence — including an explanation for any deviation from the Guidelines range." Id. at 597; ante at 189-90.

If "no significant procedural error" is identified, a reviewing court then "consider[s] the substantive reasonableness of the sentence imposed under an abuse-of-discretion standard," taking "into account the totality of the circumstances, including the extent of any variance from the Guidelines range." Gall v. United States, 128 S.Ct. at 597; ante at 189-90, 190-91. Gall emphasizes that substantive review is narrow: "it is not for the Court of Appeals to decide de novo whether the justification for a variance is sufficient or the sentence reasonable." 128 S.Ct. at 602. Rather, reviewing courts "must give due deference to the district court's decision that the § 3553(a) factors, on the whole, justify the extent of the variance." Id. at 598; see id. at 602 (same); ante at 190-91. Thus, for a sentence to be substantively unreasonable it must fall outside the "broad range" warranted by the totality of the circumstances. United States v. Jones, 531 F.3d 163, 174 (2d Cir. 2008); ante at 191.

With these principles in mind, I proceed to consider the reasonableness challenge in this case.

3. The Reasonableness Challenge at Issue

I do not understand any member of the en banc court to suggest that, absent procedural error, a 24-month sentence for illegally trafficking in 16 guns would be substantively unreasonable in this case. Thus, the singular focus of this appeal is procedural error, specifically, the district court's justification for the challenged sentence and any fact finding underlying that justification.

At the outset, I suggest that if the district court had stated simply, and without referencing any local circumstances, that upon careful consideration of the § 3553(a) factors, including the applicable Guidelines range, it concluded that a 24-month sentence was necessary (a) to reflect the seriousness of a crime that, after all, involves illegal trafficking in instruments that can take a human life; and (b) to ensure the adequate deterrence of such trafficking, we would not identify any procedural error in that justification. To demand more justification (or factual support) for such a modest variance would, in effect, challenge the district court's constitutionally mandated authority to weigh § 3553(a) factors for itself and would come close to presuming the unreasonableness of a non-Guidelines sentence. See Gall v. United States, 128 S.Ct. at 597.

The district court did not, however, presume to justify its sentence by assessing the seriousness of gun trafficking offenses generally. Rather, it more modestly considered the seriousness of Cavera's crime by reference to the community for which the illegal guns were intended: New York City. After United States v. Booker, 543 U.S. 220, 125 S.Ct. 738, 160 L.Ed.2d 621 (2005), a district court's consideration of local circumstances in assessing § 3553(a) factors cannot, by itself, manifest procedural error. This conclusion is compelled by our holding today that "we will not categorically proscribe any factor `concerning the background, character, and conduct' of the defendant, with the exception of invidious factors." Ante at 190-91 (quoting 18 U.S.C. § 3661). Accordingly, the only possible procedural challenge in this case relates to the factual basis for the district court's finding that gun smuggling to New York City has "the potential to create a substantially greater degree of harm" than in the mine-run case. United States v. Lucania, 379 F.Supp.2d at 295. See post at 212 (Straub.J., dissenting) (concluding that neither record "data" nor "reasonably available statistics" support conclusion that "New York City's population density makes Cavera's offense more serious here than in the nation generally"); post at 220 (Sotomayor, J., dissenting) (stating that "district court's analysis and data are insufficient to support its conclusion that defendant-appellant deserved a severer sentence because firearms trafficking (1) is a more serious crime in densely populated areas. . . .").

a. Kimbrough "Closer Review" Does Not Apply in This Case

Before discussing why I identify no abuse of discretion in the district court's factual assessment of the seriousness of Cavera's crime, I note my disagreement with the dissenters' view that this case warrants "closer review" in light of Kimbrough v. United States, 128 S.Ct. at 575. See post at 216-20 (Sotomayor, J., dissenting). Kimbrough referenced the possibility of "closer review" in a context not present in this case, i.e., when the Commission had relied on "empirical data and national experience" to frame a Guideline that addressed the particular concern at issue. 128 S.Ct. at 575 (internal quotation marks These circumstances were not present in Kimbrough because, although the Commission had formulated specific Guidelines to treat crack cocaine more severely than powder cocaine or heroin, it did so to mimic the drug ratios of statutes establishing mandatory minimum sentences, and not to reflect relevant empirical data or national experience. Id. (explaining why case "present[ed] no occasion for elaborative discussion" of closer review possibility).

This case is a step further removed from Kimbrough. Although the Commission has promulgated Guidelines to address the unlawful receipt, possession, or transportation of firearms, see U.S.S.G. § 2K2.1, it has never considered whether the risk of harm posed by such crimes can vary depending on the intended destination for the guns. In short, not only are there no Guidelines reflecting empirical and experiential study of this issue, there are no Guidelines at all on the point.

Some dissenting colleagues nevertheless urge "closer review" of the district court's consideration of local circumstances by construing Commission silence as the equivalent of a policy determination that local circumstances are not properly factored into a sentencing decision. see post at 217-18 (Sotomayor, J., dissenting). I disagree. Certainly, that was not how we interpreted Commission silence when the Guidelines were mandatory, and we hardly have a basis for according greater weight to Commission silence after Booker. Title 18 U.S.C. § 3553(b)(1). which before Booker mandated Guidelines sentences unless the district court identified a factor "not adequately taken into consideration by the Sentencing Commission," expressly stated that, "[i]n determining whether a circumstance was adequately taken into consideration, the court shall consider only the sentencing guidelines, policy statements, and official commentary of the Sentencing Commission" (emphasis added). We had construed this language to mean that "Congress did not limit what a court may determine under § 3553(b) to have been inadequately considered by the Commission. Rather, this provision was left open to `provide[] the flexibility necessary to assure adequate consideration of circumstances that might justify a sentence outside the guidelines.'" United States v. Lara, 905 F.2d 599, 605 (2d Cir. 1990) (quoting S.Rep. No. 98-225, at 78 (1983), reprinted in 1984 U.S.C.C.A.N. 3182, 3262). The expectation was that, as district courts identified factors not considered (or inadequately considered) by the Commission and assigned them weight in imposing individual sentences, the Commission would collect this data and study it to determine the need for new or revised Guidelines.

Thus, long before Booker, the heightened risk of harm from gun trafficking into the nation's most densely populated city had been cited as a factor not considered by the Commission that might support a (departure from Guidelines ranges. See Reena Raggi, Local Concerns, Local Insights, 5 Fed. Sent'g Rep. 306 (1993); see also Vincent L. Broderick, Local Factors in Sentencing, 5 Fed. Sent'g Rep. 314 (1993) (asserting that Sentencing Commission's silence on local conditions renders them a permissible ground for departure under 18 U.S.C. § 3553(b) as a fact not "adequately taken into consideration" in formulating the Guidelines); Charles P. Sifton, Theme and Variation: The Relationship Between National Sentencing Standards and Local Conditions, 5 Fed. Sent'g Rep. 303 (1993) (same). If the Commission had thought such a departure was unwarranted, it could have said so in a 10 Guideline or policy statement (just as it has identified family circumstances, education, age, etc., as circumstances not generally supporting departures, see U.S.S.G. §§ 5H1.1, 5H1.6). Indeed, even now, the Commission might take such action, and any resulting Guideline or policy statement would have to be considered by sentencing courts pursuant to § 3553(a)(4). But until that time, there is no legal support for us to conclude that Commission silence signals an intent to preclude judicial consideration of particular facts, and certainly no factual support to conclude that such silence is informed by the sort of empirical and experiential study that Kimbrough referenced as a basis for "closer review" of a non-Guidelines sentence.

b. The District Court's Factual Findings Do Not Manifest Clear Error

While I do not think this case calls for "closer review," whether it does or not, one thing is clear: closer review does not equate to de novo review of a sentencing court's factual findings. Such findings are reviewed only for clear error. See Gall v. United States, 128 S.Ct. at 600 (identifying error in appellate analysis that "closely resembled de novo review of the facts"); United States v. Jones, 531 F.3d at 173 (holding that "references to `closer review' and `significant justification'" do not "establish a higher standard of review than abuse of discretion" or permit review of district court's fact finding "for anything other than clear error"); cf. Maine v. Taylor, 477 U.S. 131, 144-45, 106 S.Ct. 2440, 91 L.Ed.2d 110 (1986) (observing that, even where proffered justification for discrimination is subject to strict scrutiny, "the empirical component of that scrutiny, like any other form of factfinding, is the basic responsibility of district courts" and "shall not be set aside unless clearly erroneous" (internal quotation marks omitted)).10 To identify clear error, we must do more than entertain doubts about the district court's conclusions or hypothesize reasonable alternative findings; we must be "left with the definite and firm conviction that a mistake has been committed. Anderson v. City of Bessemer City, N.C., 470 U.S. 564, 574, 105 S.Ct. 1504, 84 L.Ed.2d 518 (1985) (defining clear error) (internal quotation marks omitted); accord Maine v. Taylor, All U.S. at 145, 106 S.Ct. 2440 (applying clear error review to district court's factual findings in criminal case). The district court's finding that gun trafficking presents a greater risk of harm in New York City than in the mine-run case does not manifest such clear error.

The district court supported its finding by reference to three facts. The first two, that guns smuggled into New York City are (1) commonly acquired illegally and (2) then used for unlawful purposes, see United States v. Lucania, 379 F.Supp.2d at 295, required no empirical support. We would only have to look through our own docket history to take judicial notice of the fact that the Eastern District of New York is home to the city's five traditional organized crime families, numerous large-scale narcotics enterprises, and street gangs of all stripes — for all of whom guns are tools of the trade and gun violence a routine form of communication.

The third fact cited by the district court is New York City's status as the most densely populated area of the country. Undisputed evidence shows that the average population density of the United States is 78 people per square mile, whereas the population density in parts of New York City included in the Eastern District of New York is 35,000 per square mile. See United States v. Lucania, 379 F.Supp.2d at 295 n. 3. That yields a ratio of approximately 1:450. To be sure, if a gunman, in any part of the country, shoots at an intended victim and hits his target, the results are equally tragic. But if the gunman misses his target, the cited statistics plainly support an inference that the risk of injury, if not death, to a nearby person is many times greater in the Eastern District of New York than in most parts of the country. Similarly, if one imagines an inebriated person firing a gun into the air at random, the likelihood that it will hit someone is also far, far greater in the Eastern District of New York than in most other areas of the country. This is not to suggest that every time a gun is fired on a New York City street, at least 450 people are within range, but it shows that the number of people likely to be in range in New York is always so much higher than in many other areas of the country as to support the district court's finding of an increased risk of harm.

To the extent the district court's conclusion that trafficking guns to a densely populated city poses a heightened risk of harm is plainly grounded in common sense, our dissenting colleagues are dismissive. See post at 216 (Straub, J., dissenting) (observing that "reflexive evocation of common sense does not resolve the issue because of the countervailing evidence"); post at 219-20 (Sotomayor, J., dissenting) (noting "serious danger . . . that sentencing judges will dress their subjective views in objective trappings, either by using questionable empirical data or by invoking a `common sense' at odds with reality"). In response, I begin with a word about the role of "common sense" in judicial proceedings.

When the issue in dispute is legal — the bulk of our appellate work — rulings must plainly be grounded in law, and not ascribed simply to common sense. But, when the issue to be resolved is factual, the law expects the factfinder — whether judge or jury — to draw on common sense and experience in making any determination. See 1 Leonard B. Sand et al., Modern Federal Jury Instructions ¶ 5.02, Instr. 5-4 (2005) (noting that jury should be instructed: "You should consider the evidence in light of your own common sense and experience, and you may draw reasonable inferences from the evidence"). Indeed, common sense and experience frequently provide the necessary link between a given fact (such as the classic example of a person entering a building wearing a wet raincoat) and a permissible inference (that it is raining outside). Because appellate courts are not factfinders, when we review a challenged factual inference we do not ourselves weigh competing evidence. We ask only whether any reasonable factfinder, applying common sense and experience to the task, could have drawn the challenged inference from the record facts according to the applicable burden of proof, which at sentencing is a preponderance of the evidence. See United States v. Yannotti, 541 F.3d 112, 129 (2d Cir. 2008) (noting that the preponderance standard applies at sentence); United States v. Vaughn, 430 F.3d 518, 525 (2d Cir. 2005) (same): see also Tellabs, Inc. v. Makor Issues Rights, Ltd., 551 U.S. 308, 127 S.Ct. 2499, 2513, 168 L.Ed.2d 179 (2007) (observing that preponderance burden requires demonstration that fact at issue "is more likely than not" true). As long as the answer is yes, we defer to that finding.

In Gall, the Supreme Court emphasized the deference reviewing courts owe district courts' factual findings. In part, this is because district courts have the advantage of seeing the parties and the evidence in the particular case and assessing credibility. see Gall v. United States, 128 S.Ct. at 597-98. But this is not the only reason. Such deference also reflects an acknowledgment of the general insights and judgment that district courts develop — a sort of judicial common sense — simply by virtue of imposing scores of sentences each year. See id. at 598 n. 7 (noting that "[d]istrict judges sentence, on average, 117 defendants every year"); accord United States v. Jones, 531 F.3d at 163 n. 4. Although our dissenting colleagues are dismissive of "talismanic incantations to local `experience,'" post at 219 (Sotomayor, J., dissenting), the Supreme Court has signaled otherwise, recognizing both before and after Booker that just sentences often depend on insights drawn from the district court's "day-to-day experience in criminal sentencing," Koon v. United States, 518 U.S. 81, 98, 116 S.Ct. 2035, 135 L.Ed.2d 392 (1996), that are not always "`conveyed by the record,'" Gall v. United States, 128 S.Ct. at 597 (quoting amicus brief filed by Federal Public and Community Defenders and the National Association of Federal Defenders). While such insights may not be "quantifiable," United States v. Jones, 531 F.3d at 171 n. 4, Gall makes clear that they are an important reason why, in reviewing non-Guidelines sentences, appellate courts must "give due deference to the district court's decision that the § 3553(a) factors, on the whole, justify the extent of the variance," 128 S.Ct. at 597.

Our dissenting colleagues nevertheless insist — as a procedural matter — that more than common sense and experience was necessary to permit the district court to infer a heightened risk of harm from evidence of population density. Even before Booker, however, reviewing courts did not demand objective evidence to support Guidelines departures for extraordinary family circumstances, age, or other conditions that Commission policy statements noted were generally disfavored. See U.S.S.G. §§ 5H1.1, 5H1.6. To depart on such grounds, the district court had to find that the "discouraged factor . . . [was] present to an exceptional degree or in some other way ma[de] the case different from the ordinary case where the factor is present." Koon v. United States, 518 U.S. at 96; see also U.S.S.G. § 5K2.0; United States v. Huerta, 371 F.3d 88, 93-94 (2d Cir. 2004). Conducting such an inquiry did not require the district court to cite evidence demonstrating that the defendant's particular circumstances were "exceptional" as a matter of empirical fact. Rather, district courts were asked to "make a refined assessment of the many facts bearing on the outcome, informed by [their] vantage point and day-to-day experience in criminal sentencing." Koon v. United States, 518 U.S. at 98 (emphasis added). In light of this precedent, to condition non-15 Guidelines sentences on the production of empirical evidence would, in effect, accord a presumption of unreasonableness to such sentences, in violation of the Sixth Amendment. See Gall v. United States, 128 S.Ct. at 597; Rita v. United States, 127 S.Ct. at 2466-67.

Our dissenting colleagues submit that they have, in fact, identified empirical evidence that demonstrates clear error in the district court's finding that high population density presents a greater risk of harm from gunfire to bystanders. For example, they cite news stories reporting reductions in the number of innocent victims killed in New York City in recent years. see post at 214 (Straub, J., dissenting); post at 221 (Sotomayor, J., dissenting). Plainly, homicide is not the only form of injury that can be caused by gunfire. Indeed, the reported decline in homicides is small comfort to the significant number of New Yorkers, many of them children, who are regularly injured by random gunfire. In any event, a reduction in homicides would not be enough to identify clear error in the district court's assessment of the risk presented by random gunfire in a densely populated community. A reduction in random killings can, after all, be attributable to any number of factors, including an increased police commitment to getting illegal guns off the street, see Sworn Complaint ¶ 46, City of New York v. A-1 Jewelry Pawn, Inc., No. 06 Civ. 2233, 2006 WL 1772845 (E.D.N.Y. May 15, 2006) (alleging that 70,000 handguns were seized by the New York City Police Department from 1995-2005, "[v]irtually all . . . from individuals who were prohibited by law from possessing guns"); see also "Police Data Shows Increase in Street Stops," N.Y. Times, May 6, 2008, at Bl (reporting 145, 098 street stops by police in first three months of 2008 as part of strategy to stop crime and find illegal guns), or extraordinary community precautions that tragically confirm the danger to innocent bystanders posed by urban gun violence, see "After a Boy's Shooting, `Why?' Is on Everyone's Lips," N.Y. Times, Aug. 6, 2008, at B2 (reporting that residents of community "had learned to take careful steps to avoid getting caught in the middle of someone else's battle. They stayed indoors. They taught their children to hit the ground when a shot was heard."). Such measures, even when successful in reducing the number of times a gun is fired on New York City streets, do not alter the fact that, whenever a gun is fired, the population density of the community enhances the risk of injury beyond that of a mine-run case.

Nor is a different conclusion about the seriousness of Cavera's crime compelled by the supposed flaw that the dissenters purport to identify in the district court's reliance on a Justice Department study as evidence that "homicide rates in large urban areas remain substantially higher than in suburban and rural areas." United States v. Lucania, 379 F.Supp.2d at 295 (citing U.S. Dep't of Justice, Office of Justice Programs, Bureau of Justice Statistics, Homicide Trends in the U.S., Trends by City Size (hereafter Trends by City Size). http://www.ojp.usdoj.gov/bjs/ homicide/city.htm (last visited Nov. 24, 2008)). Our colleagues note that the cited study discusses only absolute numbers of homicides rather than relative rates of homicides. They point to a table included in the same study showing that, when the number of homicides per 100,000 inhabitants is considered, cities with populations between 250,000 and one million have recently had higher homicide rates than cities (such as New York) with one million or more residents. See post at 212-13 (Straub, J., dissenting) (citing Trends by City Size: Homicide Victimization Rates per 100,000 Population for Cities over 100,000 by Population Group, http://www.ojp.usdoj.gov/bjs/homicides/tables/vcitytab. htm (last visited Nov. 24, 2008)); see post at 220-21 (Sotomayor, J., dissenting).

This critique does not, in fact, undermine the district court's finding. The referenced table shows that the rates of homicides for larger cities, whether their populations are 250,000 to 499,000; 500,000 to 999,000; or over one million, are always significantly higher than for small cities with populations of between 100,000 and 250,000. Indeed, over the 30-year period covered by the Department of Justice's statistics, on average, larger cities had homicide rates nearly 70% higher than that of small cities ( i.e., 20.2 homicides per 100,000 versus 11.9), and cities over one million had rates nearly double that of small cities ( i.e., 23.3 homicides per 100,000 versus 11.9). Over the last five years covered by the report (2001-2005), on average, the rate in larger cities has been approximately 52% higher than that in small cities ( i.e., 14.6 versus 9.6), and approximately 45% higher in cities over one million ( i.e., 14.0 versus 9.6). On this record, the district court was not obliged to explore the reasons why cities with populations between 250,000 and 500,000 sometimes show slightly higher homicide rates than cities with populations over one million. As already noted, to the extent the explanation for this phenomenon, as far as New York City is concerned, might rest with a decade-long commitment of increased police resources to removing illegal guns from the street, that initiative would not alter the fact that unseized guns continue to pose a greater risk of harm if discharged on the streets of a densely populated city. Thus, the district court did not misstate the import of recent homicide statistics, nor do those statistics demonstrate any clear error in its ultimate finding that trafficking guns into New York City is more serious than the minerun case.

In any event, I do not understand it to be our role, at the procedural step of reasonableness review, to engage in this kind of dissection of the empirical evidence cited by the district court. Nor is it to identify competing studies or news articles pointing in other directions. As the majority opinion observes, such evidence is inevitably debatable. See ante 197 (holding that district court did not abuse its discretion in relying on potentially controversial economic theories to support findings relevant to need for deterrence). The question, at the procedural step of reasonableness review, is not whether we are persuaded to draw particular inferences from the evidence, but, rather, whether the district court committed clear error by doing so.

I identify no clear error in the district court's finding that gun trafficking to New York City is a sufficiently more serious crime than the mine-run case based on the high population density of the city as well as the likely illegal disposition and use of such guns. For that reason, as well as the heightened need for deterrence discussed in the majority opinion, I vote to uphold the challenged 24-month sentence and to affirm the judgment of conviction.

STRAUB, Circuit Judge, concurring in Parts I and II.A., and dissenting in part. Judge CARDAMONE and Judge SOTOMAYOR join in full and Judge POOLER joins in Part 3.

I join fully in my colleague's concurring and dissenting opinion. I write separately to emphasize what I believe reasonableness review entails in this case, and to explain why I believe Gerard Cavera's above-Guidelines sentence for conspiring to deal in and transport firearms, 18 U.S.C. §§ 371, 922(a)(1)(A) and (5), fails basic reasonableness review.

While it is now clear that the District Court "may vary from Guidelines ranges based solely on policy considerations, including disagreements with the Guidelines," Kimbrough v. United States, ___ U.S. ___, 128 S.Ct. 558, 570, 169 L.Ed.2d 481 (2007) (internal quotation marks and brackets omitted), it is still the case that the justification for a particular sentence must follow from the facts and premises, and those premises must be, if not ineluctable, at least sound or reasonable, see Gall v. United States, ___ U.S. ___, 128 S.Ct. 586, 597, 169 L.Ed.2d 445 (2007). The District Court's conclusions here, however, that firearms smuggled into New York City pose a greater risk of harm than in the United States generally, and that "a more severe penalty is necessary to produce adequate deterrence," do not follow from the facts. United States v. Lucania, 379 F.Supp.2d 288, 293-96 (E.D.N.Y. 2005). I appreciate the District Court's thoughtful and clearly stated consideration of the particular significance of Cavera's crime in New York City. However, viewed in light of the Supreme Court's recent guidance in Kimbrough, ___ U.S. ___, 128 S.Ct. 558, 169 L.Ed.2d 481, and Gall, ___ U.S. ___, 128 S.Ct. 586, 169 L.Ed.2d 445, the District Court's decision to enhance Cavera's sentence, as it stands, exceeded its allowable discretion. I would therefore remand this case for resentencing in light of those decisions.

We review a district court's sentence for procedural and substantive reasonableness, a standard "akin to review for abuse of discretion." United States v. Fernandez, 443 F.3d 19, 27 (2d Cir.), cert. denied, 549 U.S. 882, 127 S.Ct. 192, 166 L.Ed.2d 143 (2006); see also Gall, 128 S.Ct. at 597. Pursuant to this standard, we review a district court's interpretation of the Sentencing Guidelines de novo and apply the clear error standard when evaluating a district court's findings of fact. See United States v. Richardson, 521 F.3d 149, 156 (2d Cir. 2008).

In Kimbrough, the Supreme Court explained that sentencing "courts may vary from Guidelines ranges based solely on policy considerations, including disagreements with the Guidelines." 128 S.Ct. at 570 (internal quotation marks and brackets omitted). However, district judges do not "have a blank check to impose whatever sentences suit their fancy." United States v. Jones, 531 F.3d 163, 174 (2d Cir. 2008); see also United States v. Higdon, 531 F.3d 1561, 562 (7th Cir. 2008) (Posner, J.) (explaining that while an individual district court judge may now be free to impose "his own penal philosophy[,] . . . [a]s a matter of prudence, however, in recognition of the Commission's knowledge, experience, and staff resources, an individual judge should think long and hard before substituting his personal penal philosophy for that of the Commission"). Indeed, the Supreme Court has explained that "closer review may be in order when the sentencing judge varies from the Guidelines based solely on the judge's view that the Guidelines range fails properly to reflect [18 U.S.C] § 3553(a) considerations even in a mine-run case." Kimbrough, 128 S.Ct. at 575 (internal quotation marks omitted). And even where closer review does not apply, we have emphasized that a sentencing court must "give fair consideration to the Guidelines before imposing sentence [and] it must make `an individualized assessment' of the sentence warranted by § 3553(a) `based on the facts presented.'" Jones, 531 F.3d at 170 (quoting Gall, 128 S.Ct. at 597) (citations omitted). And we will reverse a district court's factual conclusion as clearly erroneous if it is "without foundation." United States v. Taylor, 475 F.3d 65, 70 (2d Cir. 2007) (per curiam) (internal quotation marks omitted).

The precise question presented by this case is whether the District Court exceeded its allowable discretion when it chose to rely on empirical sociological data to draw conclusions, decided that the Guidelines do not account for the judge's sociological conclusions, and then based its sentence on those conclusions. In these circumstances, I believe we must examine whether the District Court's findings are supported by the data cited in order to determine that it has a "reasoned basis for exercising [its] own legal decision-making authority." Rita v. United States, 551 U.S. 338, 127 S.Ct. 2456, 2468, 168 L.Ed.2d 203 (2007); see also Gall, 128 S.Ct. at 602 (deference is owed to sentencing court's "reasoned and reasonable decision"); United States v. Anati, 457 F.3d 233, 238 (2d Cir. 2006) (explaining, in dicta, that to the extent a judge decides "the special impact of [an] offense in a particular geographic community" is "a relevant circumstance," "there might have to be some empirical basis for deeming the impact of [that] offense in a particular community more serious than the assessment made by the Sentencing Commission"), overruled in part on other grounds by Irizarry v. United States, ___ U.S. ___, 128 S.Ct. 2198, 2201-02 n. 1, 171 L.Ed.2d 28 (2008). This, it seems to me, is what "reasonableness" review requires. It is not a question of whether the justification given is "sufficient," cf. Gall, 128 S.Ct. at 594, but of whether the justification follows from the facts and premises, and whether those premises are at least sound or reasonable. If the justification does not follow from the facts and permissible inferences, then the conclusions drawn are clearly erroneous.

The Supreme Court's analysis of the sentencing judge's use of age in Gall is instructive. In that case, the Eighth Circuit found that "the district court gave significant weight to an improper factor when it" likened Gall's sale of ecstasy at age twenty-one to the "impetuous and ill-considered actions" of persons under the age of eighteen reflected in "general studies." United States v. Gall, 446 F.3d 884, 890 (8th Cir. 2006), rev'd, ___ U.S. ___, 128 S.Ct. 586, 169 L.Ed.2d 445 (2007). The Court agreed with the Eighth Circuit that the studies on age and behavior "cited by the District Judge do not explain how Gall's specific behavior in the instant case was impetuous or ill-considered." Gall, 128 S.Ct. at 601 (internal quotation marks omitted). The Court, however, explained that the District Judge was not using the studies for that purpose, but instead was using the studies to address the defendant's character. Id. The Court then reasoned that "[g]iven the dramatic contrast between Gall's behavior before he joined the conspiracy and his conduct after withdrawing, it was not unreasonable for the District Judge to view Gall's immaturity at the time of the offense as a mitigating factor, and his later behavior as a sign that he had matured and would not engage in such impetuous and ill-considered conduct in the future." Id. As the Court's analysis makes clear, while age is not an impermissible factor, a sentencing court's conclusions about the significance of a defendant's age must be reasonable. In Gall, the sentencing judge's conclusions about the defendant's character were reasonable because they followed from the evidence about the defendant's conduct as he grew older.

Where a district court's sociological and statistical findings, as to which it enjoys no special comparative advantage vis-a-vis the Sentencing Commission, see Kimbrough, 128 S.Ct. at 574, do not reasonably follow from the evidence it cites, the court exceeds its allowable discretion and the sentence is unreasonable. In this case, in particular, two features of the District Court's analysis were unsound. I would thus find that the sentence is unreasonable and remand it to the District Court.

The District Court first erred in concluding that New York City's population density makes Cavera's offense more serious here than in the nation generally. While it is possible that this is the case, as my dissenting colleague also explains, neither the data relied on by the District Court nor reasonably available statistics support that conclusion.

The District Court cited evidence showing that New York City is the most densely populated city in the country and evidence it claimed showed that "homicide rates in large urban areas remain substantially higher than in suburban and rural areas." Lucania, 379 F.Supp.2d at 295 n. 3 (citing National League of Cities, 30 Most Densely Populated Cities, http://www.nlc.org/about_cities/cities_101/187. aspx (last visited Nov. 20, 2008); U.S. Dep't of Justice, Office of Justice Programs, Bureau of Justice Statistics, Homicide Trends in the U.S., Trends by City Size (hereinafter Trends by City Size), http://www.ojp.usdoj.gov/bjs/homicide/city.htm (last visited Nov. 20, 2008)). It then reasoned that an above-Guidelines sentence would not cause unwarranted disparities because "[f]irearms are less likely to cause harm in more rural areas, if only because they are less likely to cause harm to innocent bystanders." Lucania, 379 F.Supp.2d at 296.

But, nothing in the reports cited by the District Court supports the conclusion that the risk of harm from firearms is greater in more densely populated cities than in the rest of the country. According to the data cited, there are more homicides in large cities than in small cities and suburban and rural areas. See Trends by City Size. The report cited, however, does not compare "homicide rates" in these geographic subdivisions, as the District Court apparently believed, see Lucania, 379 F.Supp.2d at 295; rather, it compares the absolute number of homicides in large cities and other areas. As my dissenting colleague notes, when homicide victimization rates per 100,000 inhabitants are considered, the data reflect that recently homicide rates have not directly correlated with city size: cities with populations between 250,000 and one million have higher homicide rates than cities with one million or more residents. See Trends by City Size: Homicide Victimization Rates per 100,000 Population for Cities over 100,000 by Population Group, http://www.ojp. usdoj.gov/bjs/homicide/tables/vcitytab.htm (last visited Nov. 20, 2008).

When we turn to the key metric — relative rates of gun-related homicides in New York City versus elsewhere — the statistics do not support the District Court's conclusion. Homicide rates in the region that includes New York City appear to reflect the national average, while "[r]ates of murder, and especially those involving guns, are higher in southern regions of the United States — in the East South Central, West South Central, and the South Atlantic regions." U.S. Dep't of Justice, Office of Justice Programs, Bureau of Justice Statistics, Homicide Trends in the U.S., Regional Trends, http://www.ojp.usdoj. gov/bjs/homicide/region.htm (last visited Nov. 20, 2008) ("The rates of the Middle Atlantic [New Jersey, New York, Pennsylvania] and East North Central regions were closest to the national average of all regions."). In addition, in view of the fact that almost eighty percent of the United States population lives in urban areas, see U.S. Census Bureau, STATISTICAL ABSTRACT OF THE UNITED STATES: 2004-2005, Population 28, tbl.25, available at http://www.census.gov/prod/2004pubs/04statab/pop.pdf (last visited Nov. 20, 2008), the assumption that New York City is far from the national average just because most other locales are more rural, without further analysis, seems tenuous at best.

Nor, as my dissenting colleague points out, do the data suggest that innocent bystanders face a heightened risk of harm from firearms in New York City. In 2007, "[v]ery few victims of homicides were strangers to their perpetrators or were killed in random attacks." Press Release, New York Police Department, Mayor Bloomberg and Commissioner Kelly Announce that City is on course to set a New Record in Crime Reduction — Fewest Murders Since Records Have Been Kept, No. 2007-066 (Dec. 26, 2007), http://www. nyc.gov/html/nypd/html/pr/pr_2007_066.shtml (last visited Nov. 20, 2008); see also Al Baker, City Homicides Still Dropping, to Under 500, N.Y. TIMES, NOV. 23, 2007, at Al (reporting that the Police Department's official crime statistics showed in November 2007 that "with roughly half the killings analyzed, only 35 were found to be committed by strangers, a microscopic statistic in a city of more than 8.2 million"). This appears to have been the case at the time of Cavera's criminal conduct as well. Homicide statistics for 2004 reflect that "[s]treet murders are down. Innocent bystanders, once the subject of so many screaming headlines, no longer need Kevlar. . . . New Yorkers are far less likely to be killed by a stranger or casual acquaintance now than 15 years ago." Shaila K. Dewan, As Murders Fall, New Tactics Are Tried Against Remainder, N.Y. TIMES, Dec. 31, 2004, at Bl; see also Greg Gitt-rich, Death (Mostly) by Association, N.Y. DAILY NEWS, Oct. 1, 2004, at 3 (quoting Police Commissioner Raymond Kelly in 2004 as stating that "[i]t is highly unlikely that a tourist or a law-abiding citizen would be the victim of murder"). If nothing else, the statistics show that it is neither obvious nor common sense that fire-arms present a heightened risk of injury to bystanders in New York City because it is densely populated, at least not in the recent years during which Cavera committed his crime.

In sum, the fact that more homicides occur in large cities than in rural or suburban areas does not support the inference that the rate of homicides is greater in New York City than on average in this country. Even assuming a higher homicide rate in New York City, the fact that New York City is more densely populated does not support the inference that more innocent bystanders may be hurt by gun violence in New York City than on average in this country. And even assuming that more guns in New York City means more potential for harm here than on average elsewhere, there is nothing to support the assumption that trafficked guns in New York City are more likely to cause harm than they would on average in the country.

3. The District Court's Use of New York's Strict Gun Laws

The District Court also erred in concluding that "a more severe penalty is necessary to produce adequate deterrence." See Lucania, 379 F.Supp.2d at 295. The majority suggests that the District Court's deterrence rationale has "considerable support." Maj. Op. at 195-96. I do not disagree, as a general matter, that a district judge may rely on the need for greater deterrence based on a finding that fire-arms trafficking into New York City is more profitable than on average nationwide. But here, ultimately, there is no support, much less "considerable support" for the finding that firearms trafficking into New York City is more profitable than firearms trafficking on average in the country.

The article the District Court relied on to note parenthetically that New York City is "one of the `unusual areas' to which running guns is a profitable enterprise" does not make the claim that gun running into New York City is more profitable than gun running on average nationwide. See Lucania, 379 F.Supp.2d at 295 (citing Gary Kleck, BATF Gun Trace Data and the Role of Organized Gun Trafficking in Supplying Guns to Criminals, 18 ST. LOUIS. U. PUB. L.REV. 23, 41 (1999) (hereinafter Kleck, BATF Gun Trace Data)). In fact, as my dissenting colleague also explains, the study only speculates that "there may well be unusual areas, such as New York City, Washington, D.C., or Boston, where the supply of legally owned guns and stolen guns circulating among criminals is low enough to leave room for criminal entrepreneurs to make a living selling guns illegally." Kleck, BATF Gun Trace Data, at 41.

The study's main conclusion is that "organized high-volume gun trafficking appears to account for a few percent of the guns acquired by criminals," thus refuting the theory that trafficking accounts for a significant share of criminals' guns. Id. at 42 ("Criminals obtain guns . . . primarily by way of unrecorded, one-at-a-time transfers, some legal, some not, from people not in the illegal gun trafficking business. . . . [O]rganized trafficking of guns . . . accounts for no more than a tiny share of the guns obtained by criminals."). In line with that, the study undermines the notion that firearms trafficking is a significant source for criminals even in cities with strict gun laws when it notes, referring to Boston, that "even in a city subject to unusually strict gun laws, where opportunities for gun traffickers to profit should be at their maximum, probably less than 7% of crime guns recovered by police[] showed some solid indication of having been trafficked." Id. at 28-29 (emphasis added). If the vast majority of crime guns are being obtained through routes other than illegal gun trafficking, then the demand for trafficked guns and thus potential profits may not be as high as otherwise assumed. Moreover, evidence that guns cost more in New York City than elsewhere does not establish that greater profits are available in the City than in other destinations for trafficked guns. See Maj. Op. at 195-96 n. 14. Increase in the cost of a gun does not automatically lead to increased profits if the expenses also go up, say, as a result of the additional efforts needed to avoid stricter gun control enforcement.

Even assuming the evidence supports an inference that higher profits are available, the District Court did not explain how that made New York City different from other places in the country where gun trafficking occurs. It certainly bears keeping in mind that the purpose of New York State's very strict gun laws is undermined when guns are brought into the state illegally; however, that problem was the main reason for 18 U.S.C. § 922(a), and presumably its accompanying sentences, in the first place. Section 922(a)(5) was passed specifically to remedy the major problem created by persons in states with restrictive laws obtaining guns from states with less restrictive gun control laws. See, e.g., S. REP. NO. 90-1097, as reprinted in 1968 U.S.C.C.A.N. 2112, 2164 ("Two prime sources of fire-arms to criminals, juveniles, mental defectives, and crimebent individuals which involve access to guns through interstate routes are the mail-order common carrier source and the out-of-state, nonresident source. . . . Because of interstate, nonresident purchases of firearms for criminal purposes, the laws of our States and their political subdivisions are circumvented, contravened, and rendered ineffective."). Thus, as my dissenting colleague also notes, the Guidelines may already account for any deterrence rationale.

Certainly a potential for greater profits in a particular locale and a need for greater deterrence are not inconsequential matters when determining the appropriate sentence. Here, however, the conclusion that increased profits are to be had from trafficking guns into New York City does not follow from the evidence. The District Court failed to explain how the data cited supported an inference that greater profits were available to gun traffickers targeting New York City, and failed to explain how the Guidelines range inadequately accounts for the potential for greater profits given that the statute was aimed at combating that problem in the particular locales with strict gun laws.

4. Conclusion

For these reasons, I would find that the links between the facts and the conclusions in this case are so tenuous as to verge on speculation and that the sentence does not survive reasonableness review. While a need for greater deterrence and a potential greater risk to innocent bystanders are indubitably valid concerns for a sentencing judge, when a district court bases its finding that these factors are present on sociological data, there must be a reasoned basis for concluding that the data support the finding. Here, the perceived needs are not founded in the facts, the data cited do not support the inferences drawn, and reflexive evocation of common sense does not resolve the issue because of the countervailing evidence. In reviewing sentences, we would be wise to recall that "[i]t is our duty to see that the force of the state, when it is brought to bear through the sentences of our courts, is exerted with the maximum we can muster of rational thought, humanity, and compassion." Marvin E. Frankel, CRIMINAL SENTENCES: LAW WITHOUT ORDER 124 (1973). The District Court's reasoning, as it stands, leaves me wholly unconvinced that the inferences the District Court drew from the record facts and conclusions it made to support Cavera's above-Guidelines sentence were reasonable or rational. For these reasons I join my colleague's concurring and dissenting opinion. I would remand to the District Court for resentencing in light of Kimbrough, ___ U.S. ___, 128 S.Ct. 558, 169 L.Ed.2d 481, and Gall, ___ U.S. ___, 128 S.Ct. 586, 169 L.Ed.2d 445.

POOLER, Circuit Judge, dissenting:

I join in the majority's conclusions in Parts I and 11(A) regarding our authority to review sentencing decisions as well as in Judge Straub's and Sotomayor's dissents insofar as they address the economic deterrence basis for the sentence imposed below. I also agree with Judge Sotomayor's analysis of the closer standard of review that should be applied in this case. Because the majority does not rest its decision on the district's court's determination that a gun running offense is more serious when guns are transported to New York City, I take no position on that issue.

SOTOMAYOR, Circuit Judge, joined by Judge CARDAMONE and Judge STRAUB, concurs in part and dissents in part. Judge POOLER joins, in part, the dissent.

I join in the majority's conclusions in Parts I and 11(A) regarding our authority to review sentencing decisions. But I dissent from the majority's overly deferential review of the district court's variance, on general policy grounds, from the Sentencing Guidelines; and I join in my colleague's dissenting and concurring opinion. None of the district court's stated reasons for its variance were adequately supported by objective criteria. Consequently, the sentence should be vacated and the case remanded.

As the majority recognizes, the Supreme Court has held that sentencing courts, in certain circumstances, are authorized to weigh generally applicable policy factors. Maj. Op. at 191 (citing Kimbrough v. United States, ___ U.S. ___, 128 S.Ct. 558, 570, 169 L.Ed.2d 481 (2007)). The Supreme Court, however, has not suggested that this power is unfettered. To the contrary — and as acknowledged by the majority (Maj. Op. at 192 n. 9)"the Supreme Court has distinguished different categories of cases to which varying degrees of appellate scrutiny apply. Kimbrough, 128 S.Ct. at 574-75. In the case before us, "closer review" is warranted because the district court "Varie[d] from the Guidelines based solely on the judge's view that the Guidelines range fail[ed] properly to reflect [18 U.S.C.] § 3553(a) considerations even in a mine-run case." Id. at 575 (internal quotation marks omitted). The Supreme Court, however, did not elaborate on what it meant by "closer review," and the majority opinion avoids fleshing out this standard. Although its contours remain imprecise, "closer review" must amount to more than the majority's excessive deference to the district court's decision, which risks a regression of the sentencing process to the "greatest deficiencies of the pre-Guidelines regime," namely "its failure to provide for review of the decisions of sentencing judges and its failure to ensure that the sentencing judge's exercise of discretion was informed by authoritative criteria and principles." Kate Stith Jose A. Cabranes, Judging Under the Federal Sentencing Guidelines, 91 Nw. U. L.Rev. 1247, 1253-54 (1997).

Closer review is warranted where, as happened in this mine-run case, a district court implements a policy decision applicable to a wide class of offenders that is at odds with the Sentencing Commission. As the district court recognized, despite Congress's direction to consider the "community view of the gravity of the offense," 28 U.S.C. § 994(c)(4), "[t]he [Sentencing] Commission has, so far, never accepted the invitation to craft regional Guidelines," at least with respect to the firearms trafficking at issue in this case. United States v. Lucania, 379 F.Supp.2d 288, 294 (E.D.N.Y. 2005); see Reena Raggi, Local Concerns, Local Insights, 5 Fed. Sent'g Rep. 306, 306 (1993) ("When I voiced my concern to the Sentencing Commission about these guidelines for gun trafficking as they applied in New York, I was told that other parts of the country viewed gun crimes differently and that the guidelines were meant to reflect an average."). But the Sentencing Commission has deemed regional or population-based adjustments to be appropriate in other, limited circumstances. Thus, the Sentencing Commission's decision not to differentiate firearms trafficking sentences based on the fire-arms' destinations — despite a general invitation from Congress, a specific suggestion from at least one federal judge and the Sentencing Commission's consideration of population and locale in another circumstance — suggests a deliberate "balance" between "the comparative virtues and vices of broad, simple categorization and detailed, complex subcategorization." U.S. Sentencing Guidelines Manual Ch. 1 Part A.1.3.

Closer review is also appropriate because the judge's sentence, in addition to being in tension with the Sentencing Commission, was not grounded in the district court's "discrete institutional strengths." Kimbrough, 128 S.Ct. at 574; see Maj. Op. at 191-92 ("[O]ur review must be informed by the discrete institutional strengths of the Sentencing Commission and the district courts.") (internal quotation marks omitted). A sentencing judge's expertise lies in his "greater familiarity with . . . the individual case and the individual defendant before him than the [Sentencing] Commission or the appeals court." Kimbrough, 128 S.Ct. at 574 (internal quotation marks omitted). The district court's competence wanes as it moves from a case's particularities evaluated through the framework of § 3553(a) to overarching considerations of criminal jurisprudence. For example, a district court is uniquely positioned to make a refined assessment of whether a crack/powder sentencing disparity is warranted because the court sentences both defendants accused of crack-related crimes as well as defendants accused of cocaine-related crimes. See Koon v. United States, 518 U.S. 81, 98, 116 S.Ct. 2035, 135 L.Ed.2d 392 (1996) (explaining that district courts have "an institutional advantage over appellate courts" in using their "day-to-day experience in criminal sentencing" to make a "comparison with the facts of other Guidelines cases"). But the district judge is not similarly well-situated to draw comparisons between defendants in different courts around the country: A judge in Brooklyn who is evaluating the relative dangers of gun trafficking throughout the nation enjoys no institutional advantage over appellate courts or the Sentencing Commission, if only because the judge's experiences are limited to his or her region. A district court strays far from its expertise in varying from the Guidelines based on its disagreement with the Sentencing Commission — whose Congressionally mandated raison d'être is "to formulate and constantly refine national sentencing standards," Kimbrough, 128 S.Ct. at 574 — as to the proper national penal policy in response to regional differences relating to firearms trafficking.

In her concurrence, Judge Raggi argues that this case does not warrant closer review because the Sentencing Commission "has never considered whether the risk of harm posed by [firearms trafficking] crimes can vary depending on the intended destination for the guns." Concurring Op. at 202. But the impact of almost any crime will vary according to its location. See, e.g., Vincent L. Broderick, Local Factors in Sentencing, 5 Fed. Sent'g Rep. 314, 314 (1993) (arguing that "the theft of a horse would have had entirely different significance in Montana and in Manhattan"). To argue that closer review is inappropriate unless the Sentencing Commission has explicitly disfavored a particular local consideration "in a Guideline or policy statement" (Concurring Op. at 203-04) ignores the uniformity at the heart of the Guidelines. See U.S. Sentencing Guidelines Manual Ch. 1 Part A.1.3 ("Congress sought reasonable uniformity in sentencing by narrowing the wide disparity in sentences imposed for similar criminal offenses committed by similar offenders."). "Because one of the major purposes of the Guidelines was to eliminate unjustified disparities in sentences among similarly situated defendants," we should consider the Sentencing Commission to have "adequately taken a circumstance into consideration" when, as in this case, the defendant's offense "fits squarely within the language of the Guidelines" and the Sentencing Commission has made "conscious choices" regarding "the circumstances underlying the offense conduct." United States v. Stultz, 356 F.3d 261, 266-67 (2d Cir. 2004) (internal quotation marks and brackets omitted).

I believe that "closer review" means that we must test a district court's application of broad policy factors in order to ensure that the district court's conclusions can be objectively supported and are not based on faulty assumptions. This review does not amount to a policy debate. To the contrary, our review must focus on the district court's articulated reasons and cited authority, evaluating whether the latter support the former. And despite Judge Raggi's representation (Concurring Op. at 204 n. 10), such closer review does not mean that we substitute our own sentencing predilections for those of the district court. See Gall v. United States, ___ U.S. ___, 128 S.Ct. 586, 600, 169 L.Ed.2d 445 (2007) (criticizing appellate court's usurpation of district court's fact-finding role). It does, however, mean that district courts cannot immunize their decisions from appellate review by talismanic invocations to local "experience" and "a sort of judicial common sense." See Concurring Op. of J. Raggi at 205-06. For example, in Kimbrough, the district court did not rest its policy variance from the Guidelines solely upon its own courtroom experiences or legal hypotheses; instead it referenced the Sentencing Commission's extensive research and reports criticizing the crack/powder disparity. See 128 S.Ct. at 568, 575. To suggest that appellate courts should affirm a sentencing rationale if it is plausible under any set of assumptions reduces the Guidelines and appellate courts to what they are clearly not: "a body of casual advice, to be consulted or overlooked at the whim of a sentencing judge." United States v. Crosby, 397 F.3d 103, 113 (2d Cir. 2005).

As the district court recognized in this case, arbitrary and subjective considerations, such as a judge's feelings about a particular type of crime, should not form the basis of a sentence. Lucania, 379 F.Supp.2d at 296. The majority and I agree. Maj. Op. at 194-95; cf. Charles P. Sifton, Theme and Variations: The Relationship Between National Sentencing Standards and Local Conditions, 5 Fed. Sent'g Rep. 303, 303 (1993) (arguing that a sentence would create unwarranted disparity if it were motivated by "an effort to set aside national norms on the basis of local concerns without examining either the factual or legal significance of those concerns"). Yet a serious danger exists that sentencing judges will dress their subjective views in objective trappings, either by using questionable empirical data or by invoking a "common sense" at odds with reality. We only encourage this confusion if we signal that our review of sentencing decisions is cursory.

I do not suggest that the thoughtful and respected district judge in this case sentenced according to a caprice. His detailed opinion and its careful discussion of the § 3553(a) factors demonstrate otherwise. The dialogue between trial and appellate courts depends upon the candor of all judges, and the district court here is to be commended for its thorough explanation of its sentencing decision. Nevertheless, the district court's analysis and data are insufficient to support its conclusion that defendant-appellant deserved a severer sentence because firearms trafficking (1) is a more serious crime in densely populated areas, and (2) requires greater deterrence in areas with restrictive gun laws. For that reason, I believe that the enhanced sentence that the district court imposed on defendant-appellant should be vacated and remanded for reconsideration by the district court.

With respect to the seriousness of the offense, the district court relied on data compiled by the Department of Justice, which indicated that "homicide rates in large urban areas remain substantially higher than in suburban and rural areas." Lucania, 379 F.Supp.2d at 295 (citing U.S. Dep't of Justice, Office of Justice Programs, Bureau of Justice Statistics, Homicide Trends in the U.S., Trends by City Size, http://www.ojp.usdoj.gov/bjs/homicide/city.htm). Yet the data say nothing about the homicide rate in New York City, and they do not show that a gun in New York City is more likely to hurt people than a gun elsewhere. More generally, the data do not demonstrate that homicide rates increase as population density increases because they do not consider the population density of a city, but only the total population. To the extent that the data are relevant, a district court could conclude that gun trafficking into more populated cities is less dangerous than gun trafficking elsewhere because the data demonstrate that over the last five years covered by the report (2001-2005), the average homicide rate in cities with populations between 250K-499K was approximately 10% higher than the homicide rate in the largest cities. See U.S. Dep't of Justice, Office of Justice Programs, Bureau of Justice Statistics, Homicide Trends in the U.S., Homicide Victimization Rate, http://www.ojp.usdoj. gov/bjs/homicide/tables/vcitytab.htm (last visited November 19, 2008) ("DOJ Homicide Statistics"). The same data could support either a harsher or a more lenient sentence, and some of my colleagues' reasoning suggests that this Court would be required to affirm either outcome. See Concurring Op. of J. Raggi at 201 (arguing that there was no procedural error where the district court "modestly considered the seriousness of [defendant-appellant]'s crime by reference to the community for which the illegal guns were intended"). Although district courts need not always rely upon empirical data when varying from the Guidelines, a sentence should not be sustained when the data cited by a district court undermine its own conclusions.

The district court implied that common sense dictates that a gun is more dangerous in more densely populated areas because of the higher likelihood that random bystanders might be shot. Lucania, 379 F.Supp.2d at 296. But the district court erred in not scrutinizing more closely its "common sense" assumptions. Even if true that there is a slightly higher statistical probability that a gun will cause harm to a bystander in a densely populated area, it does not necessarily mean that the incremental increase in dangerousness is worthy of an increase in the sentence imposed. Moreover, the `Vast majority" of homicide victims in New York City last year were not strangers to their assailants. Al Baker, City Homicides Still Dropping, to Under 500, N.Y. Times, Nov. 23, 2007, at Al (concluding that "[t]he low number of killings by strangers belies the common imagery that New Yorkers are vulnerable to arbitrary attacks on the streets, or die in robberies that turn fatal"). This means that defendant-appellant received an increase in his sentence based on mere speculation that one of the guns he sold might have harmed a bystander in New York City, and that this harm would not have happened if he had sold the guns in a less densely populated area. The district court did not sufficiently analyze whether the risk of harm to bystanders in New York City justified a more severe sentence for defendant-appellant. See United States v. Anati 457 F.3d 233, 238 (2d Cir. 2006) (noting, in dicta, that to the extent that "the special impact of [an] offense in a particular geographic community might be relevant, there might have to be some empirical basis for deeming the impact of a heroin offense in a particular community more serious than the assessment made by the Sentencing Commission." (emphasis added)), overruled in part on other grounds by Irizarry v. United States, ___ U.S. ___ n. 1, 128 S.Ct. 2198, 2201-02 n. 1, 171 L.Ed.2d 28 (2008). Instead, it simply assumed that these weapons posed a greater danger in New York City than elsewhere because of population density.

The district court's other rationale for imposing a non-Guidelines sentence was based on "general deterrence." The majority affirms the district court's sentence on this basis (Maj. Op. at 195-97), even though the district court's reasoning was unsubstantiated and unconvincing. In concluding that firearms trafficking is more profitable in New York City than in less tightly regulated areas, the sentencing judge's sole support was a law review article that hypothesized — without the benefit of data — that "there may well be unusual areas, such as New York City, Washington, D.C., or Boston, where the supply of legally owned guns and stolen guns circulating among criminals is low enough to leave room for criminal entrepreneurs to make a living selling guns illegally." Gary Kleck, BATF Gun Trace Data and the Role of Organized Gun Trafficking in Supplying Guns to Criminals, 18 St. Louis Univ. Pub. L.Rev. 23, 41 (1999). The district judge did not make any finding as to where the majority of firearms trafficking offenses occur, and he assumed that the firearms trafficking offenses which make up the Guidelines' "national average" are spread evenly throughout the country rather than clumped in certain areas. But if the article cited by the district court is correct that the black market for guns is only profitable in a few areas like New York, Washington, D.C. and Boston, then it may well be that firearms trafficking crimes occur almost entirely or predominantly in those areas, in which case the Guidelines may already account for any deterrence issues raised by New York's strict gun laws. If so, defendant-appellant's sentence would have been needlessly increased.

The majority states that firearms trafficking is more profitable in areas with strict gun laws because more regulation increases the costs of obtaining a gun. And "[w]here the profits to be made from violating a law are higher, the penalty needs to be correspondingly higher to achieve the same amount of deterrence." Maj. Op. at 196. The majority commits a fundamental error by attempting to bolster the sentencing judge's argument by relying upon articles and economic theories never referenced by the district court. This shifts the appellate court's role from reviewing the lower court's sentencing rationale to crafting it. Even assuming that the majority's general deterrence rationale were correct, defendant-appellant's case is a particularly poor application. Defendant-appellant sold his guns in Florida, and nothing in the record indicates that he profited more on his sales to New York than on his sales to other destinations.

There are additional reasons to be skeptical and wary of the district court's theory of general deterrence. First, it is unrealistic to believe that gun traffickers willing to risk up to twenty-five years under New York law (N.Y. Penal Law §§ 70.02, 265.13 (McKinney 2008)) will now be more deterred by the possibility that, if they are charged in federal court, a federal sentencing judge in New York may increase their sentence based on the gun destination. Second, determinations regarding general deterrence may be highly subjective, and we must be careful that the individual defendant is not lost in a stereotype. If we accept the unsubstantiated general deterrence theory here, then we open the door to sentencing increases or decreases based on a litany of socio-economic factors that some study shows are linked with an increased or decreased likelihood of committing a particular crime. Finally, deterrence assumes that potential violators can anticipate what punishment they might receive. To the extent that consequences for the same federal offense vary widely from one judge to the next, deterrence is undermined because "a defendant who comes up for sentencing has no way of knowing or reliably predicting whether he will walk out of the courtroom on probation, or be locked up for a term of years that may consume the rest of his life, or something in between." Marvin E. Frankel, Criminal Sentences: Law Without Order 6 (1973); see U.S. Sentencing Guidelines Manual Ch. 1 Part A. 1.3 ("A sentencing system tailored to fit every conceivable wrinkle of each case would quickly become unworkable and seriously compromise the certainty of punishment and its deterrent effect.").

Even if I were to find adequate the district court's deterrence-based rationale for varying from the Guidelines, I still would not affirm the district court's judgment. The majority held that it need not consider the district court's argument that firearms create a greater risk of harm in more densely populated areas because the sentencing judge's deterrence rationale provided "an independently sufficient justification for its variation from the Guidelines." Maj. Op. at 195-96. But nothing in the record demonstrates that the district court would have increased the sentence to the same extent based solely on the deterrence rationale. We should not encourage district courts to compile catalogs of possible justifications for their sentences in the hope that appellate courts will sift the wheat from the chaff in pursuit of a valid reason to affirm.

I believe that district courts should be extremely careful, if not reticent, about imposing generic policy preferences on individual defendants. I do not, however, exclude the possibility that there may exist local conditions that may warrant a variance from the Guidelines if the conditions are significant and have not already been factored into the Guidelines. Our criminal justice system benefits when district courts and the Sentencing Commission engage in an "iterative process" whose outcome minimizes unwarranted sentencing disparities while appreciating individual nuances. See Tr. of Oral Arg. at 39, Gall v. United States, ___ U.S. ___, 128 S.Ct. 586, 169 L.Ed.2d 445 (Breyer, J.) ("What we want . . . is to interpret that word `reasonable' so that we get back to a situation where judges do depart when they have something unusual and maybe occasionally when they think the guideline wasn't considered properly, and then the iterative process takes over, going back to the commission. Now, how do we get there?"). But the majority's approach risks casting aside the Guidelines instead of enhancing their value.

Appellate courts must not abdicate their responsibility to ensure that sentences are based on sound judgment, lest we return to the "shameful" lack of parity, S.Rep. No. 98-225, at 65 (1983), as reprinted in 1984 U.S.C.C.A.N. 3182, 3248, which the Guidelines sought to remedy. 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. We should therefore have vacated the sentence and remanded this case to the district court for reconsideration.