Abstract
This article explores the traditional view that judges exercise broad discretion at sentencing after Booker. Around 98 percent of cases are resolved through guilty pleas, and at least 71 percent of those cases involve binding or nonbinding plea agreements, many of which stipulate to an exact sentence, guideline, or range. Parties sometimes collaborate to ensure that sentences fit within concocted United States Sentencing Guidelines ranges, and when a sentence falls within a guideline range, the U.S. Sentencing Commission never systematically collects data on the judge’s reasons for the sentence. The absence of meaningful data on judges’ reasons for two-thirds of federal sentences prevents thorough analysis of whether those sentences fulfill the intended purposes of punishment. This article contributes new data on plea agreements for sentences within guideline ranges and suggests that parties drive more of federal sentencing than previously acknowledged. Judges’ apparent complicity, particularly post-Booker, gives those sentences the cathartic veneer of Article III, maintaining a peculiar but potentially necessary framework of fictions in federal sentencing.
I. Plea Agreements and Judicial Discretion
It appears that the Sentencing Reform Act (and, more precisely, mandatory minimums, Federal Sentencing Guidelines, and the impact of the former on the latter)1 had a significant impact on the rate of guilty pleas. We can also see that Booker may have had relatively little impact on pleas overall (fig. 1). As figure 1 depicts, 97–98 percent of felonies and Class A misdemeanors are resolved by guilty pleas in today’s federal courts. On average, between 1989 and 2023, 95 percent of all sentences were imposed based on guilty pleas. But there were not plea agreements in all of those cases. Congress requires the chief judge of every district to send to the U.S. Sentencing Commission any plea agreement in every applicable case,2 and the Commission publishes how many written plea agreements it receives. If we assume that chief judges comply with this mandate, the number of written plea agreements that the Commission receives is a reasonable indicator of the number of sentences that involved written plea agreements. Figure 2 makes that comparison (though there appears to be a gap in the public data from the Sourcebook of Federal Sentencing Statistics for 1996–2002).3 On average, between 1989 and 2023, 71 percent of all cases contained written plea agreements, 24 percent did not contain plea agreements,4 and the rest were resolved by trial.
In light of the fact that almost all criminal cases are resolved by guilty pleas, and a majority of those have plea agreements, I aim to show that it is important to understand what is in those plea agreements. If binding or nonbinding plea agreements actually drove the associated sentences, any discussion of judicial discretion that ignores the central role of plea agreements is at best incomplete and at worst inaccurate or misleading.
Federal Rule of Criminal Procedure 11(c)(1)(A) (“A-Plea”) allows parties to negotiate charge-bargain agreements, 11(c)(1)(B) (“B-Plea”) allows the parties to agree on nonbinding sentencing recommendations, and 11(c)(1)(C) (“C-Plea”) allows parties to enter into binding sentencing agreements such that, if the judge rejects the agreement, the defendant may withdraw a guilty plea.5 For a host of reasons described below, knowing whether a case involved an A-, B-, or C-Plea is necessary for understanding whether judges or parties likely drove a sentence.
The problem is, outside observers do not currently know what is in most plea agreements. We often do not even know what type of Rule 11 plea agreement the parties entered into. In cases with binding plea agreements, judges defer to the parties on the sentence in some appreciable respect. The Commission has said as much and has excluded sentences based on C-Pleas from some analyses of judicial discretion, describing C-Pleas as those in which “some aspect of the sentencing was otherwise outside the court’s discretion.”6 In such cases, a judge’s individual discretion might have had little to do with the reasonableness of the sentence or which guidelines applied and more to do with trust in or deference to the parties.
The Commission collects and publishes data on sentences that are imposed based on C-Pleas, but only when the sentence deviates from the guideline range.7 The Commission derives most of the data that it codes from the Statement of Reasons form (SOR),8 and the SOR requires detailed reasons for a sentence only when the judge sentences outside of the guideline range.9 Importantly, if the sentence is within a guideline’s range, plea information is not currently tracked, outside of rare special research projects.10 While the Commission understandably cares most about when and why judges deviate from the Guidelines,11 if we care about how judges are exercising their discretion, the real reasons for a sentence matter whether or not the sentence was within a guideline range.
First, under a plain reading of 28 U.S.C. § 994(w), Congress instructed judges to state the reasons and submit documentation for all of their sentences—not just those that fall outside of the guideline range.12 Relatedly, knowing the real reasons for sentences is important if we purport to engage in real-offense sentencing.13 One pre-Booker study found that in 20–35 percent of cases parties circumvented guidelines that should have applied to the real offense.14 If parties or judges regularly circumvent the real offense’s actual guideline range through plea bargaining or other means,15 it would be a signal to the Commission or Congress that there is potentially (1) a problem with the applicable guidelines or (2) a problem of rogue parties or judges who ignore the applicable guidelines and apply ones they believe are more appropriate. But Congress and the Commission never receive this signal if it is concealed by confabulated guideline ranges (because the plea information is not reported for sentences with a guideline range). Failing to scrutinize reasons for within-range sentences also conceals potential unwarranted disparity when different offenders who commit similar conduct receive different plea deals16—undermining the entire purpose of the Guidelines.17
Knowing the true reasons for within-range sentences sheds important light on judicial discretion. For example, in the data I describe below, a judge sentenced an offender to ten years in prison for child endangerment, but most other offenders around the country with seemingly identical offense conduct received sentences of around one year. The likely intuition is to blame the judge for the disparate sentences, that is, until one learns that the longer sentence was imposed pursuant to a plea agreement that the judge accepted. The parties agreed on a sentence that was ten times higher than the national average sentence for the offense.18 To what extent is this a reflection of the judge’s sole discretion? Of course, judges retain discretion to reject plea agreements under Rule 11, but knowing more about plea agreements is at least relevant to the discussion of judicial discretion.
The same considerations are present for purportedly “lenient” sentences. The highest-profile recent example is the Supreme Court confirmation hearings for Justice Ketanji Brown Jackson, who was castigated by senators for her sentences as a trial judge in nonproduction child-pornography cases.19 But in special research projects (which predate the confirmation hearings), the Commission has found that nonproduction offenders’ sentencing exposure was reduced in 80 percent of sentences by “charging practices, guideline stipulations inconsistent with the facts recounted in [presentence reports] and/or plea agreements, government sponsored variances and departures (other than for substantial assistance), and non-government sponsored variances and departures.”20 In 80 percent of these cases, the parties tell the judges that guidelines that should technically apply are inappropriate.
There is a dearth of existing analyses on plea agreements for within-range sentences, which might stem from the difficulty in compiling a dataset.21 Because within-range sentences do not need to be explained in detail in the SOR, the only current way to piece together whether a plea agreement potentially drove a sentence is to manually review plea agreements and other (often confidential) documents and then analyze the extent to which the ultimate sentence reflects the real offense and the parties’ agreement.
In their 2023 article, Jeffrey Bellin and Jenia I. Turner found that, among cases with publicly available plea agreements, around 80 percent were A-Pleas or B-Pleas with charge bargains or sentencing recommendations (or both), and 20 percent were binding C-Pleas.22 In every case, judges accepted the plea agreements, though they appeared to deviate from the parties’ nonbinding recommendations in around a third of sentences (giving lower sentences in all but one case).23
I have additional data that can make a small contribution to the discussion. As part of a broader study, I focused on sentences imposed with and without guidelines, and I coded information about plea agreements as a part of that analysis to see whether agreements might have influenced the sentence.24 The dataset consisted of 236 cases brought under the Assimilative Crimes Act and sentenced under United States Sentencing Guideline (USSG) § 2X5.1 during fiscal years 2016–21.25
Just as Bellin and Turner found, for every sentence in my dataset that had a plea agreement, the judge accepted the plea agreement.26 Among the most common offenses in my dataset (child abuse, felony eluding, and intoxicated driving), there were written plea agreements in 91 percent of the sentences. Of these, 7 percent were A-Pleas, 27 percent B-Pleas, and 29 percent C-Pleas, and 28 percent of the time the parties either made no sentencing recommendation in the documentation or agreed to anything within the court’s determination of the ultimate guideline range. In all but two cases, the judges adopted the parties’ recommendation.27 A- and B-Pleas were likely underrepresented here. The format and content of plea agreements vary around the country, and I elected to conservatively code A-, B-, and C-Pleas only if the plea agreements expressly cited a subsection of Rule 11 or it was equally clear by the language of the agreement. Additionally, as those familiar with the practical realities of sentencing are no doubt aware, parties sometimes submit sentencing recommendations and agreements only at the sentencing hearing, which would not likely be reflected in any documentation that courts send to the Commission.28 Overall, parties (conservatively) entered into an A-, B-, or C-Plea in 63 percent of these offenses.
Compared to Bellin and Turner’s findings, the cases in my dataset had fewer total A-, B-, or C-Pleas (80% vs. 63%) but more binding C-Pleas (20% vs. 29%).29 Bellin and Turner also found that judges deviated from the parties’ proposed sentencing range more often (35% vs. only around 2%).30 A caveat, however: I do not claim that the sentences in my dataset are necessarily representative of all federal sentences. Some courts have called the types of cases in my dataset (assimilated crimes) “by definition unusual,”31 and they do not occur in all districts.32 Given the difficulty and uncertainty some courts and parties have with these cases and sentences, one might anticipate more C-Pleas as the participants seek to reduce uncertainty.
We can nevertheless use these two analyses to derive very rough insights into the extent to which the parties might drive sentences in federal courts. On average, between my conservative dataset and Bellin and Turner’s, 72 percent of cases involved an A-, B-, or C-Plea agreement. An average of around 25 percent of cases involved binding C-Pleas that were all accepted. Parties potentially drove some aspect of sentencing in over two-thirds of cases and bound the court’s discretion with a C-Plea in around a quarter of cases.33 A much more comprehensive study is needed.
II. A Framework of Fictions
While the Commission tracks basic information about the type of plea agreement when it seeks reasons for why judges sentence outside of a guideline range, my and Bellin and Turner’s inquiries are among the precious few that shed light on how plea agreements might influence judicial discretion even within a guideline range. In plenty of cases the parties agree on and courts apply the guideline that matches the real offense, but parties also enter into so-called “fictional pleas” and “hidden departures.”34 If reasons for within-range sentences were tracked, we could begin to piece this story together in a more meaningful way.
Stephen J. Schulhofer and Ilene H. Nagel’s study, published in 1996, found that parties circumvent federal sentencing guidelines in 20–35 percent of cases.35 (A comprehensive, post-Booker version of this study is needed.) Although I did not systematically track potential circumvention of guidelines in my entire dataset, I did for the most common offense in my dataset. In at least 26 percent of child-abuse cases brought in federal court, parties stipulated and judges applied what I concluded were incorrect guidelines relative to the actual offense.36 These included two sentences stipulating to simple assault when the conduct likely involved involuntary manslaughter, several cases stipulating that no guidelines applied when assault or aggravated assault guidelines should have applied, and one case of sexual contact with a minor where the parties stipulated that the guidelines should not apply (and to a five-year sentence).37
If the Commission tracked the written reasons for all sentences, trends would likely emerge. Perhaps the data would suggest that judges accept binding plea agreements at a higher rate for certain offenses, for offenders with certain criminal-history scores, for certain demographics, or based on type of representation. Correlation could exist regionally or based on caseloads per judge. The data might reveal more plea agreements for amounts just short of a certain threshold of drug quantity or monetary-loss amount. Data could show the disproportionate application of a certain guideline in relation to the frequency with which the statute(s) underlying that guideline is initially charged. In such cases, the sentence or guideline upon which the parties agree would be a red herring—we would want to focus on the actual guidelines that should have applied to the real offense that the participants sought to avoid. If every participant in the system believes that those guidelines (or statutes) are inappropriate, these might be top candidates for reform, perhaps by adjusting the base offense levels (or mandatory minimums) and then adding enhancements or reductions to capture offenders who truly belong within the original range.
Ohio Supreme Court Justice Michael Donnelly has argued for a similar type of reform. Justice Donnelly criticized fictional pleas and the circumvention of certain harsh punishments, such as decades-long requirements for sex-offender registration in unjust circumstances and mandatory minimums when a firearm is brandished during a robbery.38 He believes that these “lies[] operate as a fraud on the unsuspecting public” who are oblivious of the actual conduct that the offender committed.39 Instead of fictional pleas, “the answer is to reform the laws . . . and create diversionary programs for situations where the collateral consequences in a case justify treating the defendant differently than those in situations where such consequences do not exist.”40 Having criminal penalties that actually reflect the severity of the real offense is difficult to oppose. But calls for reform must overcome several structural fictions.
The first apparent fiction is that participants in the system actually want statutory or guideline punishments that reflect the severity of the real offense. Prosecutors are the most obvious beneficiaries of the leverage that accompanies extraordinarily harsh potential punishments, even when everyone knows that the maximums almost never apply. Courts are a corollary beneficiary of any regime that incentivizes 97 percent of cases to plead—there are simply not enough courthouses, judges, staff, and jurors for every defendant to go to trial.41 Guilty offenders have the most attenuated benefit (if any),42 but they might benefit based on the ability to plead to lesser offenses that do not accurately reflect the real offenses.43 Many participants appear to benefit at least some of the time by suspending disbelief about the real offense and its statutory punishments.
Congress engages in a related fiction. The federalization of otherwise state crimes has been criticized elsewhere,44 but Congress seems to have little interest in whether many of its criminal statutes and punishment ranges are (1) tethered to anything tangible or (2) actually enforced. One popular report estimates that there are approximately 5,199 discrete federal crimes.45 A helpful follow-up would analyze how many of these have never been or are rarely used.46 There are obviously crimes that certain members of Congress or even a majority of Congress take very seriously. “But more often than not, these federal vice laws were simply a reaction to public outcry, without regard for whether any particular federal interest was at stake or whether the states were unwilling or unable to address the problem.”47 Being seen as “tough on crime” is repeatedly a winning campaign position because “the upfront price of making penalties harsher [is] approximately zero,”48 both politically and fiscally. Unlike campaigning on controversial promises or on cutting or increasing spending, one can point to a new criminal statute or stiffer penalties for politically salient criminal conduct.
By now, hopefully, Congress is aware that, with some exceptions, there is little evidence that federalization of crime49 (as opposed to leaving it to states) or arbitrarily increasing punishments50 have meaningful deterrent effects. Congress certainly knows (because the Commission has reported this to Congress)51 that parties and judges circumvent harsh punishment ranges, only the occasional offender receives harsh statutory maximums, and many statutes demonstrably overpunish offenders. Yet Congress continues to regulate crime as a “one-way ratchet”:52 the number of crimes and severity of punishment (with few exceptions).
But comprehensiveness requires a consideration of the counterfactual: a system in which statutory punishments and guideline ranges more accurately reflect the sentiment of the participants of the system and of the public53 regarding the actual severity of offenses.54 In that system, defendants’ incentive to plead guilty would be substantially reduced, and more offenders would likely roll the dice at trial. (The credit for acceptance of responsibility55 would likely need to be substantially increased in such a system.) Either courts would become backlogged or Congress would need to create more courts.56 And if Congress took seriously the task of accurately aligning severity and punishment, the “tough-on-crime” golden goose would be spent.
Instead, a framework of fictions has emerged in the system. These fictions persist because they have proven their utility to enough of the right people, who therefore suspend disbelief. Congress continues to increase or keep on the books laws it knows will not be enforced, statutory maximums that will never be reached, and mandatory minimums it knows will be circumvented. Prosecutors threaten astronomical but ultimately fictional criminal exposure. Defendants plead guilty for less than that fictional exposure, often for conduct they didn’t commit. Judges then accept those pleas. When observers analyze sentences and potential disparity, we pretend that sentences are based on judicial discretion, overstating the role of judges and understating the role of plea agreements. And all of this evades empirical analysis, as long as the sentence is within a guideline range.
A meaningful first step toward more transparency would include adding a box or two to part IV of the SOR form to indicate whether a within-range sentence was motivated by a plea agreement (and indicate the type of agreement).57 I do not suggest that the Commission has intentionally ignored this information,58 and I assume there are reasons that it is not collected.59 But avoiding this information not only conceals existing fictions in the system but also may increase their number. Any disparity in how similarly situated offenders reach different guideline ranges is concealed, and “avoiding unwarranted sentencing disparities” was the core reason for the creation of the Guidelines and the Commission in the first place.60 At least in some cases, the Guidelines themselves, and our method of collecting information about their application, seem to enable disparities. Additionally, as long as a sentence is within the guideline range, it is presumptively (1) consistent with the Sentencing Reform Act and (2) the product of the discretion of a Senate-confirmed judge. These presumptions give within-range sentences the cathartic gloss of Article III, immunizing them from scrutiny:61A judge blessed the sentence and it’s within a guideline range—everything is fine.
Justice Antonin Scalia stated that plea bargaining is “no longer a somewhat embarrassing adjunct to our criminal justice system” and that the system would “grind to a halt” without it.62 Justice Anthony Kennedy agreed in a companion case, famously adopting the view that plea bargaining “‘is the criminal justice system.’”63Booker purported to return to judges a meaningful amount of discretion at sentencing. Twenty years of Booker invites us to ask: Is that a fiction? Is it a necessary one? Do we even want to know?
Notes
See Stephen J. Schulhofer & Ilene H. Nagel, Plea Negotiations under the Federal Sentencing Guidelines: Guideline Circumvention and Its Dynamics in the Post-Mistretta Period, 91 nw. u. l. rev. 1284, 1309 (1997) (“The drug Guidelines . . . are anchored to mandatory minimums rather than to past practice, because mandatory minimums represent a deliberate departure from past practice”).
See 18 U.S.C. § 991(c).
The Commission separates 2004 and 2005 data as before and after Booker and Blakely, respectively. These figures average the two numbers each year.
Of course, there is a chance that not all plea agreements make it into the record. See generally u.s. sentencing comm’n, fifteen years of guidelines sentencing: an assessment of how well the federal criminal justice system is achieving the goals of sentencing reform 88 (2004) [hereinafter cited as fifteen year report] (describing an attorney general memo that “reiterate[d] that plea agreements are to be placed on the record”). Additionally, in some cases the plea agreements might be oral. The Commission also notes in the Sourcebook of Federal Sentencing Statistics (1991–2023 [hereafter referred to as Sourcebook]) that it excludes an identified number of cases each year “for which the Commission was unable to determine definitively whether the defendant’s guilty plea was entered pursuant to a written plea agreement.” While this number was higher in earlier years, it has been fewer than ten cases for the past several years.
See also Jeffrey Bellin & Jenia I. Turner, Sentencing in an Era of Plea Bargains, 102 n.c. l. rev. 179, 204 (2023) (explaining types of plea agreements). Many judges dislike binding C-Pleas as unduly restrictive of their sentencing discretion. See id. However, as Bellin and Turner’s data show, C-Pleas are more common in some districts, and perhaps for certain types of cases. Of course, if a judge makes it known in their district that they are unlikely to accept such an agreement, the parties would not even try. See id. (arguing that judges’ known preferences tacitly influence plea bargaining).
See u.s. sentencing comm’n, demographic differences in sentencing: an update to the 2012 booker report 15 (2017), https://www.ussc.gov/sites/default/files/pdf/research-and-publications/research-publications/2017/20171114_Demographics.pdf.
These are currently tables 41–44 in the sourcebook. See sourcebook 73–76 (2023), https://www.ussc.gov/sites/default/files/pdf/research-and-publications/annual-reports-and-sourcebooks/2023/2023_Sourcebook.pdf (reasons for upward/downward departures/variances).
u.s. sentencing comm’n, history and purpose of statement of reasons form 15, https://www.ussc.gov/sites/default/files/pdf/training/CLE/presentation_SOR.pdf (SOR is the “primary source of information for the Sentencing Commission in coding the court’s reasons”).
Congress requires in 28 U.S.C. § 994(w) that the written SOR be made on the form issued by the Judicial Conference and approved by the Sentencing Commission. The current form is Administrative Office of the U.S. Courts Form 245B. On this form, part IV requires the judge to state whether the sentence is inside or outside of a guideline range. If the sentence is outside of the range based on a departure, the judge must then explain the reasons by filling out part V. If the sentence is outside of the range based on a variance, the judge must explain the reasons by filling out part VI. If, in contrast, the sentence is inside the guideline range (and the difference between the maximum and minimum of the range does not exceed twenty-four months), the judge must simply check box A in part IV of form 245B and can skip any parts that require a detailed explanation for the sentence.
See, e.g., fifteen year report, supra note 4 at 88 (“The only other attempt to quantify the exact impact of plea bargaining through statistical analysis was conducted by the Commission in its Four-Year Evaluation.”).
The Commission presumably relies heavily on Rita v. U.S., 551 U.S. 338 (2007), which held that for purposes of appeal, sentencing judges are not required to have given detailed reasons for within-range sentences that no party disputed. While this principle addressed the immediate constitutional issue, as stated throughout this article, the Commission and Congress have their own interests in understanding the reasons for within-guideline sentences.
28 U.S.C. § 994(w) (Chief judges in every district must submit for all cases “the written statement of reasons for the sentence imposed (which shall include the reason for any departure from the otherwise applicable guideline range)” (emphasis added)). See also 18 U.S.C. § 3553(c) (“The court, at the time of sentencing, shall state in open court the reasons for its imposition of the particular sentence,” and in addition, state reasons for sentences that are outside of the guideline range); and United States v. Pruitt, 813 F.3d 90, 93 (2d Cir. 2016) (Supreme Court precedent and 18 U.S.C. § 3553(c) “require[] sentencing judges to state the reasons for every sentence they impose”).
See u.s. sentencing comm’n, federal sentencing guidelines manual, pt. A, https://www.ussc.gov/guidelines (describing the adoption of real-offense sentencing).
See infra note 35 and accompanying text.
These have been called “fictional pleas” and “hidden departures.” See, e.g., Thea Johnson, Fictional Pleas, 94 ind. l.j. 855, 857 (2019) (a fictional plea is “a plea bargain agreement in which the defendant pleads guilty to a crime he did not commit, with the consent and knowledge of multiple actors in the criminal justice system—to avoid the profound collateral consequences that would flow from a conviction on his initial charge”); fifteen year report, supra note 4 at 82–85 (collecting authorities discussing hidden departures).
See infra note 38 (noting unwarranted disparities based on the happenstance of defense counsel negotiating better or worse plea deals for the same conduct); Adi Leibovitch, Punishing on a Curve, 111 nw. u. l. rev. 1205 (2017) (finding disparity based on the happenstance of judges’ varying dockets and familiarity with the subject matter); and u.s. sentencing comm’n, report on the continuing impact of united states v. booker on federal sentencing 97 (2012), https://www.ussc.gov/sites/default/files/pdf/news/congressional-testimony-and-reports/booker-reports/2012-booker/Part_A.pdf#page=55 (“Differences in prosecutorial practices, such as charging and plea agreement practices, have contributed to unwarranted disparity in sentencing”).
fifteen year report, supra note 4 at 11 (“The ‘first and foremost’ goal of [the Sentencing Reform Act was] avoiding unwarranted sentencing disparity,” citing Kenneth Feinberg, Federal Criminal Sentencing Reform: Congress and the United States Sentencing Commission, 28 wake forest l. rev. 291 (1993)).
It is certainly possible that reasons outside of the record justified the longer sentence, which, if true, is a problem related to the ones identified in this article.
See Ketanji Brown Jackson Confirmation Hearings, CSPAN (2022), https://www.c-span.org/search/?sdate=&edate=&congressSelect=&yearSelect=&searchtype=Videos&sort=Least+Recent+Event&text=0&tagid%5B%5D=5776&formatid%5B%5D=52.
u.s. sentencing comm’n, report to congress: federal child pornography offenses 225 (2012), https://www.ussc.gov/sites/default/files/pdf/news/congressional-testimony-and-reports/sex-offense-topics/201212-federal-child-pornography-offenses/Full_Report_to_Congress.pdf; see also u.s. sentencing comm’n, federal sentencing of child pornography: production offenses 2 (2021), https://www.ussc.gov/sites/default/files/pdf/research-and-publications/research-publications/2021/20210629_Non-Production-CP.pdf (citing a 2012 report to Congress and its findings on plea agreements).
fifteen year report, supra note 4 at 88 (stating that more research on the precise impact of presentencing stages on sentences is “sorely needed” and that extant estimates were “unreliable”).
Bellin & Turner, supra note 5 at 210.
Id.
I compiled data for that article while working as a Supreme Court Fellow in 2022–23 with a placement at the U.S. Sentencing Commission. I manually reviewed the judgment and commitment order, the SOR, any plea agreement, the indictment or other charging document, and the pre-sentence investigation report for all cases in the dataset. See Sam J. Merchant, A World without Federal Sentencing Guidelines, 102 wash. u. l. rev. 1031 (2025). Plea agreements were not the primary focus of that piece.
The Assimilative Crimes Act of 1825 provides that if conduct occurs within a federal enclave that is not otherwise punishable by federal law but is punishable by a state law in the area where the conduct occurred, the government may prosecute the state law in federal court (18 U.S.C. § 13(a)). Because these are state offenses brought in federal courts, they have no direct sentencing guidelines. USSG § 2X5.1 instructs judges that, for offenses for which there are no direct guidelines, “apply the most analogous offense guideline” and, if there is no sufficiently analogous guideline, simply impose a fair sentence under 18 U.S.C. § 3553(a).
These 100 percent acceptance rates of plea agreements could be over-representative, however. It is far from clear that there is a uniform practice of filing and then submitting to the Commission all plea agreements, even those that judges reject. If only the final, court-accepted plea agreement is filed (for analyses that derive data from the case dockets) or sent to the Commission (for analyses that derive data from documents sent to the Commission), there would be no record of rejected plea agreements.
In these two cases in the dataset, the same judge accepted the written plea agreements and recommended sentence lengths, but (correctly, in my view) rejected the stipulated guidelines because the parties analogized to inapplicable guidelines that aligned with their agreed sentence length. The judge analogized to the correct guidelines and then varied downward to reach the parties agreed sentence length.
A related study could analyze the role of recommendations made by probation officers.
See Bellin & Turner, supra note 5 at 210.
Id.
United States v. Leahy, 169 F.3d 433, 441 (7th Cir. 1999) (quotations omitted).
They occur primarily in jurisdictions with tribal land, national parks, or military bases.
Of note, the reader may want to consider that, among the 80 percent of Bellin and Turner’s dataset that were A- and B-Pleas, judges apparently varied or departed downward in one-third (35%) of these cases; see supra note 5.
See supra note 15.
Schulhofer & Nagel, supra note 1 at 1309.
This was determined by applying the “elements-based approach” to match guidelines to statutes of conviction, which is the most common approach to determine the applicable guidelines in these cases. See, e.g., United States v. Clark, 981 F.3d 1154, 1163 (10th Cir. 2020); United States v. Osborne, 164 F.3d 434 (8th Cir. 1999); and Merchant, supra note 24.
Recall that this dataset involved assimilated state offenses that have no direct guidelines, so parties and judges have some flexibility under USSG § 2X5.1 in concluding whether a guideline is “analogous” to the state offense. See Merchant, supra note 24.
Michael P. Donnelly, Truth or Consequences: Making the Case for Transparency and Reform in the Plea Negotiation Process, 17 ohio st. j. crim. l. 423, 426, 431 (2020).
Id. at 431.
Id. at 433.
Nancy J. King & Michael E. O’Neill, Appeal Waivers and the Future of Sentencing Policy, 55 duke l.j. 209, 217 (2005) (“Busy trial judges have little incentive to reject stipulations by the parties”).
For example, guilty offenders who are subject to seemingly arbitrary mandatory minimums do not benefit from these fictions. Justice Donnelly also notes the potential disparity that fictional pleas can entail when some offenders benefit from a fictional plea while others do not. See Donnelly, supra note 38 at 431 (“What about the scores of other defendants who were not so fortunate, serving decades in prison and suffering collateral consequences? How should they feel about such a resolution? Tough luck? Too bad for them? They should have hired the right lawyers?”).
Cf. John Gleeson, The Sentencing Commission and Prosecutorial Discretion: The Role of the Courts in Policing Sentence Bargains, 36 hofstra l. rev. 639, 658 (2008) (“one of the most important benefits of sentence bargains today is that they help to leaven a sentencing regime that is too harsh”).
See, e.g., Susan A. Ehrlich, The Increasing Federalization of Crime, 32 ariz. st. l.j. 825 (2000); and Kathleen F. Brickey, Criminal Mischief: The Federalization of American Criminal Law, 46 hastings l.j. 1135 (1995); but see Susan R. Klein & Ingrid B. Grobey, Debunking Claims of Over-Federalization of Criminal Law, 62 emory l.j. 1 (2012).
giancarlo canaparo et al., count the code: quantifying federalization of criminal statutes (Jan. 7, 2022), https://www.heritage.org/sites/default/files/2022-01/SR251.pdf [https://perma.cc/HM8V-RTPD].
Cf. Joel S. Johnson, Dealing with Dead Crimes, 111 geo. l.j. 95, 98 (2022) (analyzing the “pernicious effects of dead crimes,” advocating for adoption of the desuetude principle, and arguing that the continued existence of dead crimes “undermines the rule of law by enabling abuses at several stages in the criminal justice system”).
Michael A. Simons, Prosecutorial Discretion and Prosecution Guidelines: A Case Study in Controlling Federalization, 75 n.y.u. l. rev. 893, 904 (2000).
David Alan Sklansky, The Changing Political Landscape for Elected Prosecutors, 14 ohio st. j. crim. l. 647, 668 (2017).
See, e.g., Stephen Chippendale, Note, More Harm than Good: Assessing Federalization of Criminal Law, 79 minn. l. rev. 455, 467 (1994).
See, e.g., John B. Meixner Jr., Shocking Sentences, ind. l.j. 137 (2024) (collecting sources).
See, e.g., u.s. sentencing comm’n, 2016 report to the congress: career offender enhancements 2 (2016), https://www.ussc.gov/research/congressional-reports/2016-report-congress-career-offender-enhancements (recommending differentiation among predicate offenses for career offenders and finding that, “during the past ten years, the proportion of career offenders sentenced within the applicable guideline range has decreased from 43.3 percent in fiscal year 2005 to 27.5 percent in fiscal year 2014, while government sponsored departures have steadily increased from 33.9 percent to 45.6 percent”); and u.s. sentencing comm’n, report on the continuing impact of united states v. booker on federal sentencing 6 (2012), https://www.ussc.gov/sites/default/files/pdf/news/congressional-testimony-and-reports/booker-reports/2012-booker/Part_A.pdf (“For most offense types, the rate of within range sentences has decreased while the rate of below range sentences (both government sponsored and non-government sponsored) has increased over time”).
Sklansky, supra note 48 at 668.
See, e.g., Marvin E. Wolfgang et al., U.S. Dep’t of Justice, The National Survey of Crime Severity (1985).
This would take into account stacking, grouping, concurrent vs. consecutive rules, etc.
See USSG § 3E1.1 (Acceptance of Responsibility).
Alternatively, the federal government could prosecute fewer cases. Cf. Albert W. Alschuler, Plea Bargaining and Its History, 79 colum. l. rev. 1, 38 (1979) (Supreme Court cases have made criminal cases longer, more complex, and more expensive).
These boxes could simply borrow from the current options for sentences outside of the guideline range: whether the sentence was pursuant to a binding or nonbinding plea agreement or charge bargain, and if the judge deviated from the parties’ agreed sentence or range.
Retired federal judge and former Sentencing Commissioner John Gleeson suggests otherwise. Sitting by designation on the Second Circuit, he said that:
The [SOR] form as a whole seems designed to encourage judges to sentence within the range. A path of least resistance is clearly marked, and it is consistent with the Commission’s overall approach to sentencing in the post-Booker era. . . . The Sentencing Commission has since repeatedly asked Congress to enact legislation requiring sentencing courts to give greater weight to the Guidelines range than Booker and its progeny permit. . . . Thus, the objectionable part of Statement of Reasons form may reflect the law as the Commission wants it to be.
United States v. Pruitt, 813 F.3d 90, 94 n. 8 (2d Cir. 2016).
I also acknowledge that every sentencing researcher presumably has no shortage of additional information they would like to elicit on the SOR form.
18. U.S.C. § 991(b)(1)(B); and supra note 17.
Further immunizing sentences from review is the prevalence of appeal waivers in plea agreements. See, e.g., King & O’Neill, supra note 41 at 219–25, 254 (finding that, in cases with plea agreements, over two-thirds include appeal waivers; “[a]nother concern is that waiver may insulate from review sentencing issues for particular types of offenses”).
Lafler v. Cooper, 566 U.S. 156, 186 (2012) (Scalia, J., dissenting).
Missouri v. Frye, 566 U.S. 134, 144 (2012) (quoting Robert E. Scott & William J. Stuntz, Plea Bargaining as Contract, 101 yale l.j. 1909, 1912 (1992)).
