dissenting:
The majority holds that a federal sentencing enhancement should be applied under the illegal reentry guideline, U.S.S.G. § 2L1.2(b)(l)(B), whenever an offender’s rap sheet contains a prior North Carolina conviction that, given his offense and criminal history levels, could have resulted in a sentence of incarceration exceeding one year. I would hold instead that such an offense does not qualify as a predicate felony if, due to a statutorily authorized, judicially-accepted, binding plea agreement, the state sentencing judge is legally compelled to impose a sentence of no more than one year. Put differently, and consonant with our relevant circuit precedent, I would treat such an offender as if the state court judge had found him statutorily ineligible for a sentence of more than one year, which of course was true once the judge accepted his guilty plea and before imposing sentence. See infra pp. 335-37 (explaining the operation of N.C. Gen.Stat. § 15A-1023(b)).
Our disagreement as to the outcome in this case stems, I think, less over the content and application of relevant precedent and more from a fundamental disagreement regarding our role as arbiters of a flailing federal sentencing regime. Where, as here, we have been presented with a choice in how to interpret the interstices of federal sentencing law, and where one choice would exacerbate the harmful effects of over-incarceration that every cadre of social and political scientists (as well as an ever-growing cohort of elected and appointed officials, state and federal, as well as respected members of the federal judiciary) has recognized as unjust and inhumane, as well as expensive and ineffectual, this insight can and should inform our analysis. I deeply regret the panel’s failure to take advantage of the opportunity to do so here.
I.
First, some necessary and useful background.
A.
Over the latter half of the last century, enthusiasm for incarceration pervaded *331crime reduction policy and the related public discourse. The policy choices that resulted have created an unparalleled rate of incarceration — nearly 2.23 million people, or 1 out of every 100 adults, currently sit in an American prison or jail — a marked departure from the historical experience of the United States as well as the modern experience of peer democracies. Dept, of Justice, Bureau of Justice Statistics, L. Glaze & E. Herberman, Correctional Populations in the United States, 2012, at 3 (2013), http://www.bjs.gov/content/pub/pdf/ cpusl2.pdf. The United States now holds the highest prison population rate in the world, over 5 to 10 times more than western European democracies. Int’l Ctr. for Prison Studies, R. Walmsley, World Prison Population List 1-3 (10th ed.2013). Though it is home to only 5 percent of the world’s population, our nation accounts for nearly 25 percent of its prisoners. Congressional Research Service, S. Kirchhoff, Economic Impacts of Prison Growth 9 (2010), http://fas.org/sgp/crs/misc/R41177. pdf.
By all accounts, these “tough on crime” policies have been an abject failure. A rapidly accumulating group of multidisciplinary research studies have come to the conclusion that the rate of incarceration in the United States needs to be significantly reduced, and both the executive and legislative branches seem to agree. As a recent study prepared by the research arm of the National Academy of Sciences put it, the United States has “gone past the point where the numbers of people in prison can be justified by social benefits,” and arrived at a point where mass incarceration itself is a major “source of injustice.” National Research Council, J. Travis, et al., The Growth of Incarceration in the United States: Exploring Causes and Consequences 9 (2014) (“National Research Council Report”).
Justice Kennedy summarized it best ten years ago: “Our resources are misspent, our punishments too severe, our sentences too long.” Greenhouse, High Court Justice Supports Bar Plan to Ease Sentencing, N.Y. Times, June 24, 2004, p. A14. Consider the present state of our federal Bureau of Prisons: more than half of its 200,000 inmates are incarcerated for drug-related offenses, and only 6 percent for violent crimes. Dept, of Justice, Bureau of Justice Statistics, E.A. Carson & D. Goli-nelli, Prisoners in 2012: Trends in Admissions and Releases, 1991-2012, at 43 (2013), http://www.bjs.gov/content/pub/pdf/ pl2tar9112.pdf. Almost half of the inmates suffer from substance abuse disorders. Dept, of Justice, Bureau of Justice Statistics, C. Mumola & J. Karberg, Drug Use and Dependence, State and Federal Prisoners, 2004, at 1 (2006), http://www.bjs. gov/content/pub/pdf/dudsfp04.pdf. And of those released, 40 percent are rearrested or have their supervision revoked within five years, frequently for minor violations of the terms of their release. W. Rhodes, et al., Recidivism of Offenders on Federal Community Supervision 8 (2012), https:// www.ncjrs.gov/pdffilesl/bjs/grants/ 241018.pdf.
Each inmate costs our system, and thus the taxpayers, $29,291 annually. Congressional Research Service, N. James, The Federal Prison Population Buildup: Overview, Policy Changes, Issues, and Options 15 (2014), http://fas.Org/sgp/crs/misc/R 42937.pdf. A Brookings Institute project shows that direct corrections expenses total $80 billion a year; total expenditure soars to more than $260 billion once police, judicial, and legal services are included. The Hamilton Project, M. Kearney, et al., Ten Economic Facts about Crime and Incarceration in the United States 13 (2014), http://www.brookings.edu/7media/research/ files/papers/2014/05/ Olcrime-*332facts/v8_thp_10erimefacts (“Hamilton Project Report”).
Perhaps these numbers would be easier to accept if we had conclusive data that severe punishment resulted in lower crime rates. But there are no such data. “Through the 1990s and 2000s, crime rates fell significantly, but the evidence indicates it is unlikely that the rise in incarceration rates played a powerful role in this trend.” National Research Council Report, at 340. The data are, at best, mixed, and there is compelling evidence that severe prison sentences actually make reoffending more likely when offenders reenter society. Id. at 135-40, 150-52; see also, e.g., Daniel S. Nagin, Deterrence in the Twenty-First Century 201, 42 Crime and Justice 199, 201 (M.Tonry, ed. 2013) (“[Tjhere is little [empirical] evidence that increases in the length of already long prison sentences yield general deterrent effects that are sufficiently large to justify their social and economic costs.”); Anne Morrison Piehl & Bert Useem, Prisons, in Crime and Public Policy, 542 (Joan Petersilia and James Q. Wilson, eds., 2nd ed.2011) (same).
B.
The heady weight of this experiment’s failure falls disproportionately on our poor, our communities of color, and excruciatingly so on young black men:
Those who are incarcerated in U.S. prisons come largely from the most disadvantaged segments of the population. They comprise mainly [of] minority men under age 40, poorly educated, and often carrying additional deficits of drug and alcohol addiction, mental and physical illness, and a lack of work preparation or experience.... The meaning and consequences of this new reality cannot be separated from issues of social inequality and the quality of citizenship of the nation’s racial and ethnic minorities.
National Research Council Report, at 2; see also Hamilton Project Report, at 17 (“There is nearly a 70 percent chance that an African-American man without a high school diploma will be imprisoned by his mid-thirties.”). Such disparities make official responses to crime and criminality a racially fraught phenomenon. In 2011, blacks were incarcerated at nearly six times, and Hispanics at three times, the rate for non-Hispanic whites; the combination of those two groups accounted for no less than 60 percent of the total prison population. Dept, of Justice, Bureau of Justice Statistics, E.A. Carson & W. Sabol, Prisoners in 2011, at 7-8; 26 (2012), www. bjs.gov/content/pub/pdf/pl 1 .pdf.
I should note that no respected researcher has suggested that the disparities in imprisonment rates can be attributed to disparities in criminality. Studies have shown that, controlling for legally relevant differences, black defendants are more likely to be confined before trial, more likely to be sentenced to prison when non-prison sentences are available, and more likely to receive longer sentences than their white counterparts. See Michael Tonry, Punishing Race: A Continuing American Dilemma 70-76 (2011); Cassia Spohn, Racial Disparities In Prosecution, Sentencing, And Punishment 166-93, in The Oxford Handbook of Ethnicity, Crime, and Immigration (S. Bucerius, et al., ed.2013). Findings in a recent study of the New York County District Attorney’s Office by the highly-regarded Vera Institute of Justice exemplify these nationwide realities: it concluded that racial disparities manifested in nearly every identifiable point of “significant prosecutorial discretion.” See Vera Institute of Justice, B. Kutateladze & N. Andiloro, Prosecution and Racial Justice in New York County 217, http://www.vera.org/sites/default/files/ resources/downloads/ race-and-prosecu*333tion-manhattan-technical.pdf (“Vera Institute Study”).
As a result, according to a United States Sentencing Commission (USSC) report, black male offenders receive sentences 20 percent longer than those imposed on white males convicted of similar crimes. USSC, Report on the Continuing Impact of United States v. Booker on Federal Sentencing 108 (2012). The truth is that “once they’re in [the] system, people of color often face harsher punishments than their peers”; as Attorney General Eric G. Holder recently stated, “[t]his isn’t just unacceptable — it is shameful[,] unworthy of our great country, and our great legal tradition.” Eric G. Holder, Attorney General, Remarks at Annual Meeting of the American Bar Association’s House of Delegates (Aug. 12, 2013), in Justice News, http://www.justice.gov/iso/opa/ag/speeches/ 2013/ ag-speech-130812.html (“Holder Speech at ABA”) (saved as ECF opinion attachment).
C.
The dangers of over-incarceration also present themselves in the immigration context. Approximately 21,000 inmates are currently serving sentences for immigration-related offenses in the federal Bureau of Prisons. Federal Bureau of Prisons, Inmate Statistics: Offenses, http:// www.bop.gov/abouUstatistics/statistics_ inmate_offenses.jsp (last updated June 28, 2014). This sum reflects a staggering 900 percent increase in admission since 1994; in fact, immigration prosecutions now make up the single largest category of federal cases annually. Dept, of Justice, Bureau of Justice Statistics, M. Motivans, Federal Criminal Justice Trends, 2003, at 48 (2006), http://www.bjs.gov/content/pub/ pdf/fejt03.pdf; Exec. Office for U.S. Attorneys, Dep’t of Justice, United States Attorneys’ Annual Statistical Report: Fiscal Year 2012, at 10 (2012).
Immigration cases are processed in a manner bordering on mechanical. Prose-cutorial discretion is almost unheard of: less than 1 percent of immigration matters referred to U.S. Attorneys were declined for further prosecution. Dept, of Justice, Bureau of Justice Statistics, M. Motivans, Immigration Offenders in the Federal Justice System, 2010, at 18 (2013), http://www. bjs.gov/content/pub/pdf/iofjslO.pdf. Over 95 percent of immigration defendants plead guilty; in the Fourth Circuit, this figure is 98.6 percent. Id. at 8; USSC, Statistical Information Packet, Fourth Circuit, Fiscal Year 2013, at 8 (2013). The median case processing time for such cases, from inception until termination in district court, is approximately 120 days. Motivans, Immigration Offenders, at 25.
For those lacking documentation, disproportionate sentencing is still another cause for concern. Recent literature has indicated that one’s immigration status — in addition to one’s race — becomes fodder for disparate treatment at the sentencing stage. See Michael T. Light, The New Face of Legal Inequality: Noncitizens and the Long-Term Trends in Sentencing Disparities Across U.S. District Courts, 1992-2009, 48 L. & Soc. Rev. 447 (2014); see also Jeff Yates, et ah, A War on Drugs or a War on Immigrants? Expanding the Definition of “Drug Trafficking” in Determining Aggravated Felon Status for Non-citizens, 64 Md. L.Rev. 875, 880-81 (2005). According to one study, non-citizens are over three times more likely to be incarcerated compared to similarly situated citizens. Light, at 465. And the length of non-citizens’ prison terms were adversely affected, too: they were 8.5 percent longer than their counterparts, suggesting “a broader pattern of punitiveness against non-U.S. citizens, culminating in more incarceration and longer prison terms.” Id. at 466; 470.
*334All told, almost one-quarter of the Bureau of Prisons population is composed of non-citizens — over 50,000 people — and that has been the case since at least 2011. The overwhelming majority serves time for drug convictions (44%), illegal reentry (33%), illegal entry (6%), or some combination thereof, and the average sentence for this population is approximately 85 months. Bureau of Prisons Office of Public Affairs, Information on Sentenced Inmates by U.S. Citizen and non-U.S. Citizen as of Sept. 2013 (July 2014) (saved as ECF opinion attachment).* Of course, common practice for most of these non-citizen criminals is that they are passed along to be civilly deported once their federal sentences come to an end.
II.
Any reader who has come this far has my deep appreciation. Let’s examine legal doctrine.
It is against the above backdrop that we are called upon to decide the appeal of Mr. Jose Ramon Solis Valdovinos.
The facts governing Mr. Valdovinos’s appeal are not complicated: we know that in July 2009, he pled guilty in state court to four counts of selling heroin over a four-week period during the fall of 2008. Had he chosen to proceed to trial and, upon conviction, to sentencing without a binding plea agreement, Mr. Valdovinos would have faced a maximum sentence of sixteen months. N.C. Gen.Stat. § 15A-1340.17(c) & (d) (2008). But we also know that, upon his actual conviction, the state sentencing court ceased to have the option of sentencing him in that range. Because the state prosecutor had offered a plea agreement, and because the state judge unconditionally approved its terms, including the binding provision to impose a sentence of no more than twelve months, the court could only sentence Mr. Valdovinos to a determinate range of ten to twelve months’ imprisonment. J.A. 80. And so it did. Id.
The underlying legal issue is also straightforward. Though we are guided by Supreme Court holdings in Carachuri-Rosendo v. Holder, 560 U.S. 563, 130 S.Ct. 2577, 177 L.Ed.2d 68 (2010), and United States v. Rodriquez, 553 U.S. 377, 128 S.Ct. 1783, 170 L.Ed.2d 719 (2008), as well as our own precedent in Simmons and its progeny, see infra, my colleagues and I are in agreement that this case presents a novel issue of federal sentencing law, one that is posed only by virtue of the special circumstances presented by prior convictions under North Carolina’s Structured Sentencing Act and obtained pursuant to a related state statute authorizing a certain kind of plea agreement. The particular question is: which characteristic of Mr. Valdovinos’s 2009 North Carolina conviction, (1) the terms of his judicially-accepted, and therefore statutorily-binding, non-felony (for federal law purposes) sentence, imposed by virtue of a duly-negotiated binding plea agreement, or (2) the sentence that could have been imposed on a defendant with his offense class and criminal history category who goes to trial or pleads guilty without the benefit of a binding plea agreement, should govern the analysis of whether he has committed a prior “felony ... offense” for purposes of § 2L1.2(b)(1)(B), the federal sentencing enhancement applicable to those convicted of illegal reentry after deportation.
The majority chooses the latter option, largely on the ground that the “principles animating [the] decision in Simmons,” ante, at 327, support such a conclusion. Perhaps so. I choose the former, and my reasoning is as follows: (1) it is wholly consistent with the lessons of Carachuri-Rosendo, Rodriquez, and Simmons, and their “animating principles,” not least *335among them a due regard for federalism interests, and (2) in the absence of precedent mandating a result, our decision ought to be grounded in an informed understanding of the realities of the existing state and federal sentencing regimes and the consequences that our rulings may bring to bear.
A.
Let us begin with the Supreme Court’s instructions in Carachuri-Rosendo and Rodriquez. Both cases involved the use of a defendant’s prior state conviction to justify the later imposition of enhanced penalties under federal law, and both teach a single lesson: “[W]e are to look to the conviction itself as our starting place, not to what might have or could have been charged.” Carachuri-Rosendo, 560 U.S. at 576, 130 S.Ct. 2577 (emphasis added). In Carachuri-Rosendo, the Court eschewed the so-called “hypothetical” approach, which would have permitted federal law to treat the defendant as having committed an aggravated felony if, hypothetically, his previous state court proceedings could have treated him as such. This method, according to the Court, inappropriately ignored both “the conviction (the relevant statutory hook), and the conduct actually punished by the state offense.” Id. at 580, 130 S.Ct. 2577.
We later observed, in United States v. Simmons, that a state court finding could “set the maximum term of imprisonment, but only when the finding [of recidivism] is a part of the record of conviction.” 649 F.3d 237, 243 (4th Cir.2011) (en banc) (quoting Carachuri-Rosendo, 560 U.S. at 577, n. 12, 130 S.Ct. 2577) (emphasis added). And “in those cases in which the records that may properly be consulted do not show that the defendant faced the possibility of a recidivist enhancement, it may well be that the Government will be precluded from establishing that a conviction was for a qualifying offense.” Rodriquez, 553 U.S. at 389, 128 S.Ct. 1783. Use of “facts outside the record of conviction ... cannot and does not” substantiate a conclusion to the contrary. Simmons, 649 F.3d at 244 (internal citations omitted).
No one doubts that a sentence of twelve months does not qualify as a felony sentence under federal law. (The majority opinion elides Mr. Valdovinos’s careful differentiation between the definitions of “felony” for state and federal purposes, as it elides, as well, his differentiation between “felons” and “aggravated offenders.” Ante, at 328-29.) If we look, as Simmons instructs, to Mr. Valdovinos’s record of conviction, it is clear that his conviction was not “punishable by imprisonment for a term exceeding one year” as is required by the Sentencing Guidelines. U.S.S.G. § 2L1.2 cmt. n. 2. According to the state court judgment, the court “impose[d] the prison term pursuant to a plea arrangement as to sentence under Article 58 of G.S. Chapter 15A.” J.A. 80. The court had accepted Mr. Valdovinos’s guilty plea, and, under North Carolina law, it was obligated to impose the sentence agreed to by the prosecutor and defendant, which, in this case, was ten to twelve months.
Mr. Valdovinos’s appeal underscores a crucial characteristic of the plea negotiation system as enacted by the North Carolina legislature: once the plea agreement is “approved,” Article 58 of G.S. Chapter 15A “establish[es the] maximum term of imprisonment” that can legally be imposed on a particular defendant. Simmons, 649 F.3d at 244. Specifically, under North Carolina law,
Before accepting a plea pursuant to a plea arrangement in which the prosecutor has agreed to recommend a particular sentence, the judge must advise the parties whether he approves the arrangement and will dispose of the case accordingly. If the judge rejects the *336arrangement, he must so inform the parties, refuse to accept the defendant’s plea of guilty or no contest, and advise the defendant personally that neither the State nor the defendant is bound by the rejected arrangement.
N.C. Gen.Stat. § 15A-1023(b) (emphasis added). Indeed, that is the entire point of the system: once a judge accepts a 1023(b) guilty plea, she is bound by the terms of the corresponding plea agreement, and she may not go on to rewrite its terms in a manner she sees fít. Cf. Freeman v. United States, — U.S. -, 131 S.Ct. 2685, 2696, 180 L.Ed.2d 519 (2011) (Sotomayor, J., concurring) (observing that the “very purpose” of Fed.R.Crim.P. 11(c)(1)(C), the federal analogue of a North Carolina 1023(b) plea agreement, is “to bind the district court and allow the Government and the defendant to determine what sentence he will receive.”). In other words, the North Carolina judge was not “guide[d]” by the terms of the plea agreement; rather, the specific sentence is “mandate[d]” by North Carolina statutory law. Cf. Simmons, 649 F.3d at 244.
If the above language sounds familiar, this is because we pointed to this precise characteristic to support our conclusion in Simmons that the North Carolina Structured Sentencing Act should inform our federal predicate felony analysis. There, we observed that the North Carolina legislature had set forth a rigid procedure that made sentencing ranges “strictly contingent” on a defendant’s offense class and prior record level, and expressly limited sentences above that range “unless the judge makes written findings” on the record. Id. at 240. Unlike the federal Guidelines system, under which the sentencing judge could impose a sentence outside of the suggested range, a North Carolina judge lacked such discretion. To read into a state conviction a finding of aggravation that no judge ever made, and that (as in the instant case) is beyond the legal authority of the sentencing judge to make at all, is to use “facts outside the record of conviction” in a manner barred by Carachuri-Rosendo. Id. at 244-45.
Though the government boldly takes the position that some enactments of the North Carolina legislature are more important than others, this argument is unavailing. Through Article 58 of G.S. Chapter 15A, the North Carolina legislature has implemented a rigid procedure that makes sentencing ranges “strictly contingent” on the agreed-upon plea agreement — and in fact allows for no option for the judge to sentence a defendant outside that range. Mr. Valdovinos’s “record of conviction” makes clear that the maximum possible term of imprisonment was the range set forth in the plea agreement, and, as to this “conviction itself,” the state court was compelled to impose a sentence no greater than one year. We should take Carachu-ri-Rosendo and Simmons at their word and decline the government’s invitation to pick and choose among subsisting enactments of the North Carolina legislature, assigning to such enactments tiers of importance or creating statutory hierarchies that have no basis whatsoever in federal sentencing law.
The majority disagrees. By the “conviction itself,” it hastens to explain, Simmons actually meant the “offense of conviction” itself, and by the “offense of conviction” itself, it really meant an offender’s maximum punishment given his offense class and criminal history point. My friends contend that Mr. Valdovinos’s argument does not make sense because it imparts undue importance to the “moment of conviction,” which is not, after all, some sort of “magical” moment. Ante, at 327.
The “moment of conviction” may not be magical (little if anything in our broken criminal justice system is, notwithstanding its “considerable virtues” extolled by some, *337see Hon. J. Harvie Wilkinson III, In Defense of American Criminal Justice, 67 Yand. L.Rev. 1099, 1172 (2014)), but it certainly is a logical one, an apt benchmark to apply in the course of exercising our discretion to make choices about mass and prolonged incarceration. And though it is painfully obvious to say so, there simply is no “conviction itself,” see Carachuri-Rosendo, 560 U.S. at 576, 130 S.Ct. 2577, until the “moment” of conviction, and there can be no “record of conviction,” see Simmons, 649 F.3d at 243, until the judge “approves” the plea agreement and accepts the guilty plea. “Any bargain between the parties is contingent until the court accepts the agreement.” Freeman, 131 S.Ct. at 2692 (plurality op.); see also id., 131 S.Ct. at 2696 (Sotomayor, J., concurring) (treating as important, if not “magical,” “the moment of sentencing”). Unless and until the guilty plea is accepted, the state of affairs surrounding a North Carolina defendant with the benefit of a 1023(b) plea agreement is speculative at best — and certainly just as “hypothetical” as the circumstances, addressed in Simmons and Rodriquez, of a person who stands charged with offense conduct that could have exposed him to an aggravated sentencing range but the prosecutor and/or judge declined to pursue that course. If hypotheticals are inoperative in the latter circumstances, then they should be inoperative in the former.
5{5 ‡ % # % í¡í
The critical point here is that neither the holding nor the reasoning of Simmons mandates the proper resolution of this case. Despite the protestation from the majority to the contrary, ante, at 330, Car-achuri-Rosendo and Simmons are not “controlling” here, in that their “animating principles” do not compel the result reached by the majority. (Of course, on the other hand, there is a reason why our opinions are labeled “majority” and “dissent:” they have the votes, but not the better approach or the better arguments.)
Whether or not the majority will acknowledge it, we have a choice in fashioning our rule of decision here, as appellate judges sometimes do. What might inform our choice in this instance?
B.
How about the important federalism interests at stake in this case?
That is, in the course of offering the plea agreement at issue, the state of North Carolina evaluated Mr. Valdovinos’s background and the circumstances of the case and determined that he deserved a sentence of ten to twelve months-or, more aptly put, that he did not deserve a sentence of any greater duration than twelve months. This decision was made by the state’s local prosecutors, whom we presume to have their fingers on the pulse of community concerns and to act with genuine regard for community mores. Their decision-making authority is of course validated by the state Constitution and statutes, as well as its corresponding rules of procedure. Today’s holding goes far to derogate the discretion and independence exercised by state officials to enforce their own laws.
Both Carachuri-Rosendo and Simmons expressed concern that applying hypothetical sentencing enhancements to prior state convictions triggers significant federalism concerns. Carachuri-Rosendo emphasized that the federal sentencing regime should not “denigrate the independent judgment of state prosecutors to execute the laws” of their states. 560 U.S. at 580, 130 S.Ct. 2577. And we have reiterated that when a state prosecutor has “declined to pursue” a defendant as an aggravated offender, we ought not to “second-guess” her judgment. Simmons, 649 F.3d at 249-50.
*338The same logic applies here — in fact, even more so, insofar as it involves the independent but collective exercise of the combined constitutional authority of both the state prosecutor and the state judge, each expressing the community’s judgment that Mr. Valdovinos is not among the most dangerous and incorrigible offenders deserving of the full retributive weight allowed under state law.
For any number of reasons, be it pragmatism, compassion, or otherwise, the state prosecutor (and state judge) weighed the recourses available to her and opted to agree to an incarceration term of no more than one year. What today’s holding says is that that doesn’t matter: federal courts can and will “second-guess” this judgment. Even though a state prosecutor (and state judge) chose a shorter, more humane term of imprisonment, we are urged to disregard her decision because it does not comport with the policy choice of one United States Attorney’s Office as to its view of state sentencing values.
The government concedes, as it must, that “a guilty plea entered into [sic] under § 15A-1023(b) restricts a sentencing judge’s discretion,” Gov. Br. 19, but the majority concludes that this does not matter. It points to the fact that each plea agreement is individualized, hinging on “the ability of the parties to reach a deal,” ante, at 328. And it emphasizes the fact that Mr. Valdovinos “chose to plead guilty under a plea agreement that allowed him to avoid trial,” ante, at- 329. Manifestly, this describes evei^y plea agreement, all of which also “allow” the prosecutor to “avoid trial.” The majority’s reasoning is aimed at identifying a seeming contrast, I suppose, to the Structured Sentencing Act’s more wide-lensed “reflection of] North Carolina’s judgment as to the seriousness of a North Carolina crime.” Ante, at 328. But its reasoning proves too much.
First, the majority’s assertion that “a plea agreement reflects only the interests of the prosecutor and individual defendant,” id., would surely strike many as shockingly ill-informed. Every federal district judge in this circuit knows that plea agreements in the federal system are subjected to rigorous review for conformity to broad policies by multiple levels of supervisory prosecutors, whose initials customarily appear in the margins of the written agreements. Cf. Vera Institute Study, at 115-16 (describing Manhattan District Attorney’s guidelines for plea agreement offers). There is no reason to suppose, as the majority opinion seems to suggest, that conscientious prosecutors in a jurisdiction as large as Mecklenburg County, North Carolina are any less rigorous in fulfilling their responsibilities to the public. In my view, such discretionary exercises of state authority are equally instructive — if not more so — of “North Carolina’s judgment as to the seriousness” of a criminal offense committed within its jurisdiction and in violation of its own law.
Indeed, the fact that the North Carolina legislature has instituted such a plea agreement system in the first instance belies the majority’s dismissive approach. As every prosecutor and criminal practitioner well knows, a plea agreement that binds a judge to a particular sentence is a horse of a different color, for most judges will not routinely bind themselves. That the special procedure is embodied in a duly-enacted state statute undoubtedly heightens the respect we owe it. I simply fail to see how the particularized evaluation of the need for just punishment by a local prosecutor (an agent of a duly-elected, Constitutional officer of the sovereign State of North Carolina), under the authority of state statutory law, of the actual facts at issue, and agreed to by a state judge (likewise, a duly-elected, Constitutional officer of the sovereign State of *339North Carolina), can be dismissed so blithely.
And while it is certainly true that Mr. Valdovinos “chose to plead guilty under a plea agreement that allowed him to avoid trial[,]” ante, at 329, the benefits afforded to the prosecutor when a plea agreement is accepted are equally individualized and equally critical to the administration of her office’s duties. “[T]he reality [is] that criminal justice today is for the most part a system of pleas, not a system of trials”. Lafler v. Cooper, — U.S. -, 132 S.Ct. 1376, 1388, 182 L.Ed.2d 398 (2012); see also id. (over 97% of federal convictions and 94% of state convictions are a result of guilty pleas); USSC, Statistical Information Packet, Fourth Circuit, Fiscal Year 2013, at 8 (2013) (guilty pleas resolve 98.6% of immigration cases in the Fourth Circuit). It is true that the presiding prosecutor offers plea agreements for any number of reasons: a weak case, a sympathetic defendant, the expense of trial, and on and on. As, a matter of course, however, she will only reach a plea agreement if it is in the state’s interest (i.e., the community’s interest) to do so.
Actually, the majority’s ostensible distinction between the sentencing act and the plea agreement statute seems to distract more than it informs. There is no doubt that any plea negotiation system is, by its very nature, flexible and individualized, grounded in real world intentions and consequences. But there is also no doubt that it plays a crucial role in North Carolina’s sentencing scheme. Though its instructions cannot be fashioned into a neat table or grid, it strictly and steadfastly “mandates [the] specific sentences” available to the state court when the court accepts a prosecutor’s recommended sentence for any given conviction. Cf. Simmons, 649 F.3d at 244. At least according to Simmons, that is what should carry the day.
III.
My point thus far has been that the lessons of Carachuri-Rosendo and Rodriquez are consistent with acceptance of Mr. Valdovinos’s argument, and that the principles animating Simmons remain equally at play in this case, as well. In other words, as is common in cases involving the intricacies of federal sentencing law, our traditional tools do not provide us with a clearly mandated holding. I do not believe that my good colleagues in the majority dispute this; rather, where we differ is what we choose to do with this jurisprudential license.
The truly baffling question is why, when presented with a choice in the interpretation of federal sentencing law, any federal appellate judge acting in good faith (as my friends in the majority indisputably are) would choose to exacerbate, rather than mitigate, the harmful effects of our nation’s “throw-away-the-key” approach to incarceration?
Now, more than ever, every measure of political and social scientists has recognized that our nation’s mass incarceration strategies have been a “moral, legal, social, and economic disaster” that “cannot end soon enough.” Editorial, End Mass Incarceration Now, N.Y. Times, May 24, 2014, p. SR10. In February of this year, the U.S. House of Representatives renewed the bipartisan task force it created to review the federal criminal code and the trend toward “over-criminalization;” groups who have testified in support of reform include the American Bar Association, the Heritage Foundation, and, just this past month, the Judicial Conference of the United States and the Sentencing Commission. See Hearings Before the Over-Criminalization Task Force of 2013 of the Committee on the Judiciary, House of Representatives (June 14, 2013) (Testimony of William N. Shepherd, American Bar Association; *340John G. Malcolm, The Heritage Foundation); id. (Jul. 11, 2014) (Testimony of Hon. Irene M. Keeley, Judicial Conference of the United States; Hon. Patti B. Saris, USSC).
The consensus for reform includes, not least of all, the Attorney General himself, who has concluded that “far too many Americans [are] serving too much time in too many prisons — and beyond the point of serving any good law enforcement reason.” Eric G. Holder, Attorney General, Remarks at Fourth Meeting of Ministers Responsible for Public Security in the Americas (Nov. 21, 2013), in Justice News, http://www.justice.gov/iso/opa/agfepeeches/ 2013/ag-speech-131121.html (saved as ECF opinion attachment). The Department of Justice has directed federal prosecutors to exercise their discretion toward minimizing the number of inmates in federal prisons for low-level drug crimes and has urged Congress to enact changes in the federal sentencing guidelines to that effect. See U.S. Department of Justice, Smart on Crime: Reforming the Criminal Justice System for the 21st Century (2013).
In fact, even as I write this, the United States Sentencing Commission has issued a momentous, unanimous decision providing that its previously-approved Guidelines amendment reducing federal drug base offense levels by two would be retroactively applied. While Congress retains the authority to reject this twenty-first century innovation before the end of this year, it seems highly unlikely that it will do so, for all the reasons discussed herein. It is, indeed, nearly impossible to keep up with the groundswell of support evidencing our long-overdue recognition that federal sentencing law and policy are in desperate need of repair.
All of this gives me pause. We federal judges have invested no uncertain effort into crafting our tools of legal analysis, and on many occasions those traditional tools reveal a true and worthy answer. But at a point where actors from all sides of the political spectrum have concluded that federal sentencing law and policies have gone off the rails, at a point where even the Executive Branch has recognized that “widespread incarceration at the federal, state, and local levels is both ineffective and unsustainable,” see Holder Speech at ABA, I would think that our analytical process ought not to blink at these very real concerns.
The majority opinion declines to examine Mr. Valdovinos’s case through this lens, clinging instead to the myopic notion that only our ostensibly “legal” analytical tools dictate the holding in his case. I understand, in some instances, the need for formalist thought and decision-making. But in the context of federal sentencing, and in the face of mounting evidence of the societal costs of this sort of legal reasoning, I cannot condone or join in it.
I suppose, in truth, this case is really United States v. Kerr, 737 F.3d 33 (4th Cir.2013), redux. (Kerr earns a single citation in the majority opinion, see ante, at 329, but it looms, ominously, over this appeal.) Just as the North Carolina state judge there had discretion to sentence in the presumptive range (but did not), so too, here, the North Carolina state judge had discretion to reject the binding plea agreement (but did not). In each instance, reliance on an inchoate, hypothetical state of affairs to lengthen a federal sentence runs into the teeth of the relevant circuit precedents. I dissented in Kerr and I do so here. Formalist, counter-factual responses to real world events hold no comfort for me when the subject is federal sentencing.
IV.
I am reminded of the following prudent instruction from the National Academy of Sciences:
*341The decision to deprive another human being of his or her liberty is, at root, anchored in beliefs about the just relationship between the individual and society and the role of criminal sanctions in preserving the social compact.' Thus, good justice policy is necessarily based on a combination of empirical research and explicit normative commitments.
National Research Council Report, at 341. Where there are choices that can be made that would permit progress in the individual case without doing harm to the transcendent legal infrastructure rooted in deductive reasoning, we can and should choose that path.
Here, in a tiny corner of the chaotic morass that is federal sentencing law, Mr. Valdovinos has offered us a measured approach, to a novel issue of federal sentencing law, that adheres to Supreme Court and our relevant circuit precedents and is consistent with our values. If accepted by this panel, his argument, which is surely more than merely “clever”, see ante, at 326, would affect a tiny number of federal cases drawing legal relevance from North Carolina’s historical (and now superseded) sentencing regime. And Mr. Valdovinos’s sentence in this case likely would be reduced to a bottom guideline of 15 months, instead of the bottom guideline sentence he received, 27 months. He’d soon be on his way home to Mexico, if not already arrived.
That the majority declines the opportunity to decide this case on the foundations discussed herein is regrettable, a choice that not only ignores the growing wisdom informed by widespread acknowledgement of our unjust federal sentencing jurisprudence, but actually hinders its progress. Would that my friends could see that it’s a new century, complete with a host of profound and valuable insights at our avail. I discern no compelling reason why, in the performance of our adjudicative responsibilities, which every member of the panel has unfailingly carried out to the best of our ability in this case and in full accordance with our solemn oath to “administer justice,” 28 U.S.C. § 453, we ought not to draw on these insights.
One of them is that sometimes, in our shared quest for justice under law, it requires so little of us to achieve so much.
Respectfully, I dissent.
These data were provided directly to me, upon request, by the Bureau of Prisons.