United States v. VanDam

F I L E D United States Court of Appeals Tenth Circuit PU BL ISH July 10, 2007 UNITED STATES COURT O F APPEALS Elisabeth A. Shumaker Clerk of Court TENTH CIRCUIT U N ITED STA TES O F A M ER ICA, Plaintiff-Appellee, v. No. 06-4104 LEONARD DALE VANDAM , also know n as Sebastian Lopez, also known as Sebastian Lopez M endoza, Defendant-Appellant. Appeal from the U nited States District Court for the District of Utah (No. 1:01-CR-101-DB) Howard A. Pincus, Assistant Federal Public Defender (Raymond P. M oore, Federal Public Defender, with him on the briefs), Denver, Colorado, for the Defendant – Appellant. Diana Hagen, Office of the United States Attorney, District of Utah (Brett L. Tolman, United States Attorney, Stephen J. Sorenson, Assistant United States Attorney, with her on the brief), Salt Lake City, Utah, for the Plaintiff – Appellee. Before L UC ER O, M cKA Y, and HO LM ES, Circuit Judges. HO LM ES, Circuit Judge. Defendant Leonard Dale V anDam argues that the government breached its duty under his plea agreement by failing to recommend a term of imprisonment at the bottom of the applicable range under the United States Sentencing Guidelines (the “Guidelines”). As a remedy, M r. VanDam seeks specific performance before the same judge. W e agree that the government breached the plea agreement. W e also agree on these facts that M r. V anDam should receive his requested remedy. Accordingly, exercising our jurisdiction under 18 U.S.C. § 3742(a)(1) and 28 U.S.C. § 1291, we VAC ATE M r. VanDam's sentence and REM AND to the same sentencing judge for resentencing. I. BACKGROUND On December 19, 2001, M r. VanDam w as charged with one count of possession of methamphetamine with intent to distribute, in violation of 21 U.S.C. § 841(a)(1), and with one count of possession of a firearm as a felon, in violation of 18 U.S.C. § 922(g)(2). The government filed a superseding indictment on April 24, 2002. The superseding indictment added a second drug trafficking charge. The two drug trafficking charges stemmed, respectively, from incidents on October 17, 2001 and February 25, 2002. The state police arrested M r. V anDam on each occasion. The firearm charge was linked only to the October 17, 2001 incident. 2 A. Plea Agreement After M r. VanDam unsuccessfully moved to suppress the evidence seized in connection with the October 17, 2001 incident, he entered into a conditional plea agreement with the government. M r. VanDam pleaded guilty to the October 17, 2001 drug trafficking charge and to the felon-in-possession charge, although he reserved his right to appeal the suppression ruling. In exchange, the government dropped the February 25, 2002 drug-trafficking charge and, inter alia, agreed to “recommend sentencing at the low end of the guideline range found applicable.” R., Supp. Vol. II, Doc. No. 48, ¶ 11(C), at 4. The district court accepted M r. V anDam’s guilty plea on December 13, 2002. B. First Sentencing The Presentence Investigation Report (“PSR”) recommended a Guidelines range of 188 to 235 months of imprisonment, based upon a total offense level of 33. In calculating the base offense level pursuant to Guidelines § 2D1.1, the PSR converted the cash found in M r. VanDam's possession on October 17, 2001 into a quantity of methamphetamine and then into a marijuana equivalent. At sentencing, the district court chose not to convert the cash attributed to M r. VanD am into a drug equivalent for purposes of calculating his base offense level. W ithout the conversion, the district court found M r. VanDam's total offense level to be 31, rather than 33. This yielded a Guidelines range of 151 to 188 months. The district court then chose to sentence M r. VanDam in the middle 3 of that range, to a 168-month period of incarceration. C. First Appeal M r. VanDam appealed the denial of his motion to suppress. On appeal, the government conceded that the district court should have suppressed a firearm found in M r. VanDam's vehicle. This Court agreed that the firearm was illegally obtained. 1 See United States v. VanDam, No. 03-4137, slip op. at 1 (10th Cir. Sept. 15, 2005). W e therefore vacated M r. VanDam's original sentence and remanded for additional proceedings. Id. at 2. D. Second Sentencing On remand, the parties chose to honor the original plea agreement, without the felon-in-possession charge. Prior to resentencing, the government filed a sentencing memorandum. The government argued that the offense level – and, therefore, the Guidelines range – should be higher than that found applicable during the original sentencing. The government contended that possession of the firearm still triggered an enhancement under Guidelines § 2D1.1(b)(1), since the exclusionary rule is generally inapplicable to sentencing proceedings. The government also argued that the district court erred during the first sentencing proceeding in not 1 W e rejected the district court's theory that the search of M r. VanDam's car was valid because the parking lot in which his car was located functioned as curtilage to the motel room, the search of which was supported by probable cause. 4 converting the cash in M r. VanDam's possession at the time of his October 17, 2001 arrest into a drug quantity. Asking the district court to consider the firearm and the cash, the government pushed for a Guidelines range of 188 to 235 months. The government advocated for the “low end of that guideline range.” R., Supp. Vol. II, D oc. 84, at 4 (emphasis added). The government's sentencing memorandum did recognize, however, that the district court previously rejected this offense-level calculation. Faced with this reality, but stressing that the applicable Guidelines range should be 188 to 235 months, the government conceded that the district court’s original sentence of 168 months was still a “reasonable sentence.” Id. A sentencing hearing was conducted on April 25, 2006. At the outset, the district court summarized the government's position as to the appropriate sentence: The government claims that . . . all of this effort by M r. VanDam and all that reading and you successfully getting this remanded does not make any difference. They still want the 168 months. In fact, they think it really should be 188 months, but they are willing to live with 168. Right? R., Vol. II, Tr. at 5. The government responded, “That is exactly our position, yes.” Id. (emphasis added). Later, the government elaborated upon this position: If the Court had counted that money, we would end up 5 with a final offense level of 33 and a criminal history of four for a range of 188 to 235 months. Discounting the money, his final offense level is down to a level 31 with a range of 151 to 188. The 168 is within that guideline range. The government initially agreed to recommend the low end of the guideline range as a part of the sentencing considerations in this case. However, the government is of the opinion that the appropriate guideline range would be that 188 to 235, and the low end of that would be 188. Id. at 7-8 (emphasis added). At the end of its colloquy, the government presented its “bottom line”: it was “still recommending the 168 month sentence be imposed at this resentence.” Id. at 9. Following the government's ultimate recommendation of 168 months, M r. VanDam personally addressed the district court. He described his personal progress since his first sentencing. In particular, M r. VanDam noted that, due to the absence of any disciplinary problems, he was transferred from a medium- security facility to a low-security facility; that he was promoted in his prison job; and that he was taking yoga and Spanish classes. M r. VanDam also took responsibility and expressed contrition for his past offenses. After these statements, the district court announced M r. VanDam's new sentence. The district court rejected the government's Guidelines calculation, concluding that the original offense level of 31 was still “appropriate.” Id. at 14. This yielded a range of 151 to 188 months. Unlike the first sentencing, however, the district court sentenced M r. VanDam at the low end of this range, to 151 months. It also credited M r. VanDam for the 15 months that he had spent in 6 pretrial detention on federal and state charges, based upon its belief that the U.S. Bureau of Prisons would refuse to do so. The district court rested its more lenient sentence – its movement from the middle of the G uidelines range to the low end of the G uidelines range – upon M r. VanDam's strides towards rehabilitation. It explained that “M r. VanDam is doing remarkably well in prison and following the rules and actually finding what he claims to be some new balance in life and hopefully a disassociation from the use of methamphetamine and the practices that got him in trouble in the first place.” Id. at 14. The district court did not mention the government's actual recommendation of 168 months in this analysis. Nor did the district court refer to the government's obligation under the plea agreement to recommend the low end of the applicable Guidelines range. After announcing M r. VanDam's prison sentence, the district court considered his request to change one of the conditions of his supervised release, which w ould require him to maintain full-time employment. The district court conversed with M r. V anDam about this request, and ultimately granted it. During this conversation, the government agreed that it had no objection and otherw ise remained silent. E. Instant Appeal On M ay 1, 2006, M r. VanDam filed a timely notice of appeal. M r. VanDam argues that: (1) the government breached the plea agreement by not 7 recommending a sentence at the low end of the Guidelines range and that, as a remedy, he is entitled to specific performance of the plea agreement before the same sentencing judge; and (2) the district court erred by not applying the exclusionary rule at sentencing. II. D ISC USSIO N The viable part of M r. VanDam's appeal relates to the alleged breach of the plea agreement. 2 R esolving this appeal requires a three-step analysis: (1) we must determine whether the government breached the plea agreement; (2) if so, we must determine whether the breach can be deemed harmless; and (3) if the breach cannot be deemed harmless, we must determine what remedy M r. VanDam should receive. W e first hold that the government breached the plea agreement by not recommending a sentence at the low end of the Guidelines range that the district court found to be applicable – that is, a range of 151 to 181 months. W e then 2 M r. VanDam raises the exclusionary-rule argument “only to preserve it for consideration by the full court and/or by the Supreme Court.” Aplt Br. at 27. M r. VanDam concedes that this Panel must adhere to United States v. Ryan, 236 F.3d 1268 (10th Cir. 2001), which held that evidence obtained through an illegal search and seizure may be introduced at sentencing, unless the evidence was illegally gathered with the intent to secure an increased sentence. Id. at 1272. M r. VanDam acknowledges the inapplicability of the Ryan exception. Accordingly, because this Panel may not overrule a decision of another Panel absent en banc reconsideration or a supervening Supreme Court decision, see United States v. M eyers, 200 F.3d 715, 720 (10th Cir. 2000), we apply Ryan and summarily reject M r. VanDam's contention that the district court erred by not applying the exclusionary rule at sentencing. 8 hold that, pursuant to Santobello v. New York, 404 U.S. 257 (1971), and its progeny, it is inappropriate to apply a harmless-error analysis. Finally, we hold that M r. VanDam should receive the remedy of specific performance before the same sentencing judge. A. Breach This Court reviews de novo the question of whether the government has breached a plea agreement, even when the defendant fails to preserve this objection below. See United States v. Werner, 317 F.3d 1168, 1169 (10th Cir. 2003). General principles of contract law define the content and scope of the government's obligations under a plea agreement. See, e.g., United States v. Guzman, 318 F.3d 1191, 1195 (10th Cir. 2003). W e thus look to the express language in the agreement to identify both the nature of the government's promise and the defendant's reasonable understanding of this promise at the time of the entry of the guilty plea. See, e.g., United States v. Rockwell Int'l Corp., 124 F.3d 1194, 1199 (10th Cir. 1997). W e construe all ambiguities against the government, to the extent it is the drafting party. See, e.g., United States v. Brye, 146 F.3d 1207, 1210 (10th Cir. 1998). W e evaluate the record as a whole to ascertain whether the government complied with its promise. See United States v. Rodriguez-D elm a, 456 F.3d 1246, 1251 (10th Cir. 2006) (concluding after “consideration of the entire record” that government did not breach plea 9 agreement), cert. denied, 127 S. Ct. 1338 (2007). Under the unambiguous language of the plea agreement in this case, the government was obligated to recommend a sentence “at the low end of the guideline range found applicable.” R., Supp. Vol. II, Doc. 48, ¶ 11(C), at 4. This language did not preclude the government from advocating for a particular Guidelines range. However, it did require the government to recommend a sentence at the bottom of the range that the district court ultimately “found applicable,” even if that range differed from the government's proposed range. Both the government and M r. VanDam agree with this interpretation of the plea agreement. See Aplee. Br. at 13; A plt. Br. at 18. Reading the record as a whole, the government breached the plea agreement. At no point in its written submissions or oral statements did the government recommend a sentence at the bottom of the ultimately-applicable Guidelines range, as defined by the district court – i.e., 151 to 188 months. Instead, the government recommended a sentence at the low end of the range it deemed applicable. This combination of inaction and action, respectively, undermined the government's promise to M r. V anDam. See United States v. Cachucha, 484 F.3d 1266, 1270 (10th Cir. 2007) (holding that prosecutor breached plea-agreement promise “not to seek a higher offense level or an upward variance”; his statements disapproving applicable Guidelines range “undermine[d] the prosecutor’s promises” to defendant). 10 W e start with the sentencing memorandum. In this document, the government stated that M r. VanDam's Guidelines range should be 188 to 235 months. It then recommended a sentence at “the low end of that guideline range.” R., Supp. Vol. II, Doc. 84, at 4 (emphasis added). The government's sentencing memorandum ultimately conceded the reasonableness of a specific sentence that was below the range that the government believed to be applicable – that is, 168 months. Nonetheless, this fixed 168-month recommendation was in the middle, not the low end, of the range that the district court found to be applicable (i.e., 151 to 188 months). At the sentencing hearing, the government again confirmed that M r. VanDam's sentence “should be 188 months,” although it was “willing to live with 168 [months].” R., Vol. II, Tr. at 5. W hen it did mention its sentence- recommendation obligation under the plea agreement, the government did so in a manner that suggested the obligation was no longer in effect: Discounting the money, his final offense level is down to a level 31, with a range of 151 to 188. The 168 is within that guideline range. The government initially agreed to recommend the low end of the guideline range as a part of the sentencing considerations in this case. However, the government is of the opinion that the appropriate guideline range would be that 188 to 235, and the low end of that would be 188. Id. at 7 (emphasis added). In fact, these comments suggested that the government felt obligated only to recommend a sentence at the low end of its preferred 11 Guidelines range. The government further projected this belief in its last remarks to the district court on the sentencing issue. The government expressed a willingness to accept the district court’s original sentence, which was below the Guidelines range favored by the government but 17 months above the low end of the Guidelines range the district court ultimately “found applicable”: “I guess the bottom line is we are still recommending the 168 month sentence be imposed at this resentence.” Id. at 9 (emphasis added). Put simply, neither the 188-month recommendation nor the 168-month recommendation was at the bottom of the range “found applicable” by the district court. Yet, they were the only recommendations the government offered. Both directly contravened the government’s obligations under the plea agreement. The government correctly points out that the PSR informed the district court of the government's obligation to “recommend the defendant be sentenced at the low-end of the applicable guideline range.” R., Vol. III, ¶ 6 at 3. The government argues that, pursuant to United States v. Smith, 140 F.3d 1325, 1327 (10th Cir. 1998), the PSR fulfilled its duty under the plea agreement. W e disagree. In Smith, we held that “the term 'recommendation' in a plea agreement does not require the prosecutor to allocute in favor of specific adjustments in the defendant's sentence if the recommendations are contained in the PSR and the prosecutor does not allocute against an agreed-upon adjustment.” 140 F.3d at 12 1327. Applying this interpretation, we concluded that the government did not breach its obligation to recommend various sentencing adjustments even though it never made an argument for any of the adjustments at the sentencing proceeding. Id. at 1327. W e observed that the sentencing court was aware of the government's promises, which were memorialized in the PSR; they “were noted by the judge at the sentencing hearing.” Id. And, importantly, “[n]o act of the government before or at sentencing could be read as contrary to the position they agreed to support.” Id. at 1327-28. However, the circumstances identified in Smith that would permit the recitals of a PSR, alone, to satisfy a governmental promise to make a particular sentence recommendation are not present in this case. Unlike what transpired in Smith, the government in this case did not just remain silent at the sentencing hearing. Instead, it advocated for a sentence contrary to the agreed-upon recommendation. Furthermore, unlike Smith, the sentencing court in this case never acknowledged at the hearing the PSR's references to the government's plea- agreement recommendations. Accordingly, the PSR’s mere recital of the government's plea-agreement obligations in this case in no way relieved the government of the responsibility for honoring those obligations through its actions. The government notes, however, that the district court acknowledged the government's obligations under the plea agreement at the original change-of-plea 13 hearing, in December 2002. However, the second sentencing took place more than three years after the original change-of-plea hearing. It therefore is unlikely, to say the least, that in resentencing M r. VanDam the district court recalled its prior statement concerning the government’s plea-agreement obligations. See United States v. Riggs, 287 F.3d 221, 223-25 (1st Cir. 2002) (holding that government breached plea agreement by failing to recommend sentence based upon specific drug quantity, even though sentencing judge acknowledged sentencing recommendation that government was obligated to make at Rule 11 hearing six months earlier). 3 Finally, the government argues that, even if it needed to make additional comm ents recommending the low end of the ultimately-applicable Guidelines range, it had no opportunity to do so because the district court identified this range and then swiftly pronounced M r. VanDam's sentence. W e reject this type of rationalization. The government had many opportunities to comply with the plea agreement. It could have made the district court aware of its obligation prior to the sentencing hearing, perhaps through the government's sentencing memorandum. During its sentencing-hearing colloquy, the government 3 W e take no position on whether close temporal proximity between a Rule 11 colloquy, during which a district court recites the government's sentence- recommendation obligations, and a sentencing hearing is enough, without more, to satisfy the government's duty under a plea agreement to recommend a particular sentence. 14 unambiguously could have informed the district court that, although it wanted a certain range, it nonetheless was recommending the low end of whatever range the district court found to be appropriate. Indeed, the government also could have honored its agreement by respectfully getting the district court’s attention and interjecting the requisite recommendation immediately after the district court identified the applicable Guidelines range but prior to the oral pronouncement of the sentence. None of these steps was taken. 4 4 The lack-of-time argument is belied further by the fact that the district court employed the 151- to 188-month Guidelines range at the original sentencing hearing. Thus, as it admitted during oral argument, the government was on notice throughout the resentencing process that the district court could use this range at the resentencing hearing. Accordingly, the government was w ell situated at the hearing (as w ell as before) to craft arguments that took into account the possibility that the court would find to be applicable a lower-than- preferred Guidelines range. To be sure, the government insisted during oral argument that M r. VanDam’s failure to object to its arguments for a higher sentencing range in connection with the resentencing effectively made the 188- to 235-month range the operative baseline for the resentencing – a baseline from which the district court deviated only at the very end of the hearing. W e are unpersuaded. Irrespective of whether M r. VanDam objected to its sentencing arguments, the government had a plea-agreement obligation to make clear to the district court that it recommended the low end of w hatever G uidelines range the district court found to be applicable; the government failed to do so. Furthermore, given that the district court had specifically rejected the 188- to 235-month range at the original sentencing, the government would have been operating in a truly Panglossian fashion if it chose to believe that its arguments (unopposed or otherwise) had the power to reimpose this higher range as the resentencing status quo. Indeed, the district court apparently held the obverse view regarding the operative sentencing baseline. In other words, the record strongly suggests that the district court viewed the 151- to 188-month range to be the status quo, from (continued...) 15 Based upon the record as whole, the government never recommended a sentence at the low end of the applicable Guidelines range, as found by the district court. Rather, the government consistently recommended sentences either below or at the low end of the range it thought should apply. The government therefore breached the plea agreement. B. H arm lessness The government argues that even if a breach took place, the breach itself was harmless because the district court imposed a sentence in conformity with what the government should have recommended – a sentence at the low end of the Guidelines range. To support this position, the government claims that the Supreme Court's holding in Santobello, which did not apply a harmless error analysis to a prosecutorial breach of a plea agreement, is inapposite because M r. VanDam “seeks only re-sentencing and expressly disavows the position that his plea should be vacated.” Aplee. Br. at 20. W e believe that Santobello is not only relevant, but also dispositive: it precludes harmless-error review. In Santobello, the Supreme Court answered tw o necessarily sequential questions: (1) the antecedent question of whether relief is 4 (...continued) which it declined to deviate to reach the higher range that the government urged upon it. The district court stated: “So I’m not going to change the offense level, with all due respect to you [prosecutor]. . . .” R., Vol. II, Tr. at 14 (emphasis added). In sum, the government’s lack-of-time argument is without merit. 16 required for a breach of a plea agreement; and (2) the subsequent question of what remedy to provide. See 404 U.S. at 262-63. The Supreme Court answered the first question in the affirmative, at least with respect to a material breach. 5 See id. at 262 (noting that breach was not “immaterial”). It identified two principles that require automatic remand, regardless of whether the government's material breach of the plea agreement influenced the sentencing judge: (1) promoting the “interests of justice”; and (2) recognizing “the duties of the prosecution in relation to [guilty-plea] promises.” 5 Santobello provides no express guidance as to the effect of non- material breaches. In one instance, after “find[ing] [that] the breach by the Government was not material,” we held that no relief was warranted. United States v. Easterling, 921 F.2d 1073, 1080 (10th Cir. 1990). W e did not apply harmless-error review in Easterling. Instead, we scrutinized the unique factual record, which evidenced, inter alia, the district court’s approval of the defendant’s request to enforce a plea agreement that spoke to his cooperation with law enforcement. Id. W e then determined that “any breach by the Government” in failing to submit a timely cooperation memorandum to the district court that would have only further elaborated upon the defendant’s law enforcement cooperation was “immaterial.” Id. In a similar vein, some courts have addressed such situations through what might be called the “technical violation” doctrine. This doctrine holds that “technical” or “minor” violations of plea agreements (i.e., de m inimis breaches) do not require a remedy, absent evidence of actual prejudice at a sentencing hearing. See United States v. Amico, 416 F.3d 163, 167-68 (2d Cir. 2005); United States v. Clark, 55 F.3d 9, 14 n.3 (1st Cir. 1995). The rationale behind this doctrine is that only material breaches of a plea agreement, as in this case, cause a defendant's reasonable contractual expectations to go unfulfilled. See United States v. Vaval, 404 F.3d 144, 155 (2d Cir. 2005). At least in express terms, w e have not previously endorsed the technical violation doctrine. Under the facts of this case, which involves a material plea-agreement breach, we have no occasion to do so. 17 Id. In essence, the Supreme Court concluded that due to the gravity of these interests, which transcend a particular sentencing outcome in a particular case, the judiciary must provide some “relief.” Id. at 263. This determination in no way hinged upon the type of remedy the defendant requested (e.g., specific performance at a resentencing or plea-agreement vacatur). After announcing that “relief” was required, the Supreme Court then identified the available remedies – the lesser remedy of specific performance or the greater remedy of withdrawal of the guilty plea. Id. The Supreme Court gave the sentencing court the discretion to choose between these two options, based upon “the circumstances of this case.” Id. at 263. M oreover, in stating that resentencing, if chosen, “should” take place before a different judge, the Supreme Court recognized the need for the appearance of institutional fairness. Id. A harmless-error analysis violates the principles behind Santobello's automatic-remand rule. Due to the institutional nature of these principles, the government's material breach of a plea agreement cannot be rendered harmless simply because a defendant seeks the remedy of resentencing, rather than vacatur of the plea agreement. See United States v. Vaval, 404 F.3d 144, 155 (2d Cir. 2005) (“[I]n order to preserve the integrity of plea bargaining procedures and public confidence in the criminal justice system, a defendant is generally entitled to the enforcement of a plea agreement without showing a tangible harm resulting from a breach.”). 18 For instance, the “interests of justice” rationale implicates, at a minimum, principles of fundamental fairness. Clearly, it is inequitable to permit a prosecutor to renege on the consideration that induced a defendant to waive important constitutional guarantees. This inequity – not getting the benefit of the constitutional bargain (i.e., government action/inaction and the potential to influence the sentencing judge) – invalidates the agreement, including the knowing and voluntary waiver. See Mabry v. Johnson, 467 U.S. 504, 509 (1984) (“It follows that when the prosecution breaches its promise with respect to an executed plea agreement, the defendant pleads guilty on a false premise, and hence his conviction cannot stand.”). A defendant therefore is entitled to a remedy that restores the meaningfulness of his bargain, see Santobello, 404 U.S. at 262 (noting that agreement “must be fulfilled” to maintain integrity of plea), even if an appellate court convinces itself beyond a reasonable doubt that the trial judge w ould impose the same sentence on remand. See Dunn v. Colleran, 247 F.3d 450, 461-62 (3d Cir. 2001) (“Especially when the prosecutor's promise is not binding on the court, the defendant does not bargain for a specific sentence but for a lock on what the prosecutor can do and say at sentencing. That the sentencing court does not follow the prosecutor's lead is irrelevant.”). Santobello's second justification for the automatic-remand rule – “appropriate recognition of the duties of the prosecution” – militates even m ore 19 vigorously against a harmless-error analysis. A prosecutor's material breach of a plea agreement undermines the efficacy, and probity, of the criminal justice system. It also erodes public confidence in the legitimacy of this system. A finding of harmlessness – that the breach had no effect on a particular sentencing court – fails to cure these structural wounds. It does nothing to make the system whole. In other words, the necessity of preserving the integrity of the criminal justice process, and the public's faith in this integrity, requires correction of any material governmental breach, regardless of the sentencing impact of that breach on a particular defendant and the form of the ultimate remedy. See United States v. M ondragon, 228 F.3d 978, 981 (9th Cir. 2000) (“The harmless error rule does not apply when the government breaches a plea agreement. The integrity of our judicial system requires that the government strictly comply with its obligations under a plea agreement.” (emphasis added and citations omitted)). There is nothing in Santobello to suggest that a material breach can be rendered harmless. The Court never once intimated that in certain situations relief may not be appropriate. N or did it suggest that, if the remedy of specific performance was chosen on remand, the underlying breach might be transformed into an innocuous one. In fact, Santobello only addressed the question of the nature of the remedy because the impact of the breach itself could not be deemed 20 harmless. 6 W e have adopted, at least implicitly, this reading of Santobello – as holding that considerations of fairness and the integrity of the judicial system render harmful any material breach, regardless of the remedy requested. See United States v. Hawley, 93 F.3d 682, 693-94 (10th Cir. 1996) (holding that defendant necessarily is entitled to relief from material breach of plea agreement “regardless of whether the government's conduct actually affected the sentencing judge”). This reading also has been expressly embraced by virtually every circuit that has addressed the question of whether to apply harmless error. See Vaval, 404 F.3d at 154-56; M ondragon, 228 F.3d at 981; United States v. Askew, 1999 W L 236187, at *2 (6th Cir. April 14, 1999) (unpublished); United States v. Clark, 55 F.3d 9, 14 (1st Cir. 1995); United States v. M artin, 788 F.2d 184, 187 (3d Cir. 1986); see also Correale v. United States, 479 F.2d 944, 949 (1st Cir. 1973). But cf. United 6 The government's proposed approach – determining whether to apply harmless-error review to a material breach after identifying the remedy to rectify the government's error – ignores the independent impact of that breach. The concurring and dissenting opinion's reasoning suffers from the same failing. See C& D Op. at 1, 3 (Lucero, J., concurring and dissenting). This methodology incorrectly hinges the availability of harmless-error review upon the particular remedy that a defendant chooses. It evaluates not the effect of the harm but, rather, the effectiveness of the desired remedy; that is, its effectiveness in obtaining an objective that in fact cannot be guaranteed through the plea bargaining process (i.e., a lesser sentence). Consequently, this methodology overlooks, and carries the potential to leave uncorrected, the harm to the integrity of the plea and to the criminal justice system as a whole. This is the very result that Santobello's automatic-remand rule sought to avoid. 21 States v. William s, 656 F.2d 357, 359 (8th Cir. 1981) (noting Santobello rule requiring relief but finding breach harmless because district court “effectively granted specific performance of the plea agreement” through Rule 35 proceeding). Animating these cases is a recognition that a finding of harmlessness is impossible because the institutional harm caused by a prosecutor's material breach transcends any prejudice that a single defendant might suffer from a particular sentencing outcome. See, e.g., Vaval, 404 F.3d at 154-55; see also Dunn, 247 F.3d at 461-62 (describing rationale for inapplicability of harmless-error review). As in Santobello, these courts reached a conclusion as to the impossibility of harmlessness prior to addressing the question of the appropriate remedy. See Vaval, 404 F.3d at 156; Clark, 55 F.3d at 14-15; M artin, 788 F.2d at 187; Correale, 479 F.2d at 949. Significantly, many of these decisions found a harmless-error analysis inappropriate, even though the courts ultimately ordered the remedy of specific performance. See Vaval, 404 F.3d at 156 (ordering specific performance); M ondragon, 228 F.3d at 981 (same); Clark, 55 F.3d at 15 (same); Correale, 479 F.2d at 949-50 (same). In ordering specific performance at a resentencing, these courts in virtually every instance necessarily accepted the possibility that the sentence upon remand would be the same. 7 W ithout question, their logic applies 7 To be sure, these decisions did not involve a situation where a (continued...) 22 with equal force to the instant appeal: a material breach is harmful on a institutional level, regardless of whether a defendant, who seeks resentencing, would receive the same sentence absent the breach. See, e.g., Clark, 55 F.3d at 14 (“That the district court did not consider the government’s breaching sentencing memorandum is irrelevant to the question of whether Clark is entitled to a remedy.”). Our decision in United States v. Belt, 89 F.3d 710 (10th Cir. 1996), does not advance the government’s cause. In Belt, we carved out a legal-impossibility exception to Santobello's automatic-remand rule. Id. at 713. Specifically, a prosecutorial breach of a plea agreement can be harmless when a defendant seeks the remedy of resentencing and it is legally impossible for the sentencing outcome upon remand to be different. Id. Applying this exception, we held that the government's breach of its duty not to make a sentence recommendation was harmless because the defendant sought only the remedy of specific performance and because the original sentence was the lowest sentence authorized by law. Id. 7 (...continued) defendant – due to his request – might end up being sentenced in front of the same judge. That is a somew hat unusual circumstance. Under Santobello, as discussed further infra, the typical course is to remand to a different judge, when the remedy will be resentencing. See, e.g., Cachucha, 484 F.3d at 1271. Arguably, a defendant may be more likely to receive the same sentence if the judge is the same. However, Santobello has made clear that an identity of sentencing outcomes is irrelevant to the question of whether relief is appropriate for a material, prosecutorial breach of a plea agreement. And, as discussed in text, that holding is supported by very strong institutional concerns. 23 W e stressed that the applicability of harmless-error review was confined to “the very limited context presented [t]here.” Id. The Belt exception to Santobello's automatic-remand rule is therefore a legal anomaly. On these facts, Belt does not speak to the appropriateness of harmless-error review. Although M r. VanDam seeks the remedy of resentencing, a lower sentence is legally possible. By its terms, then, Belt is inapposite. As the government correctly points out, we did not expressly state in Belt that the circumstances present there “are the only circumstances in which the breach of a plea agreement may be found harmless.” Aplee. Br. at 21. Given the profound institutional concerns animating the holding of Santobello, however, w e are hard pressed to conceive of any other circumstances that would justify application of harmless-error review. In summary, the government's proposed harmless-error analysis contradicts Santobello and the policy concerns that animated its automatic-remand rule. Because the government's material breach was necessarily harmful, we must now determine the appropriate remedy. C. Remedy M r. VanDam does not seek to vacate his guilty plea. Instead, he requests that the government be required to fulfill its promise in front of the same sentencing judge. Although the choice of remedy rests with the court rather than the 24 defendant, see Allen v. Hadden, 57 F.3d 1529, 1534 (10th Cir. 1995), we are inclined to grant M r. VanDam the relief that he requests. Resentencing is the typical remedy for the breach of a plea agreement. E.g., Brye, 146 F.3d at 1213 (describing two available remedies and noting that defendant may only withdraw guilty plea when breach is “particularly egregious or intentional”); see United States v. Camper, 66 F.3d 229, 232 (9th Cir. 1995). W hen, as in this case, the defendant does not seek to withdraw his guilty plea, the less drastic remedy of resentencing appears to be most apt. See Cachucha, 484 F.3d at 1271. W e also conclude on these facts that it is appropriate to permit M r. VanDam to be sentenced before the same judge, despite the general rule that resentencing should take place in front of a different judge. 8 See Santobello, 404 8 W e in no way endorse the position that remand to the same judge is the usual remedy. The District of Columbia Circuit has adopted this view. See, e.g., United States v. Wolff, 127 F.3d 84, 88 (D .C. Cir. 1997); see also United States v. Bowler, 585 F.2d 851, 856 (7th Cir. 1978) (finding under facts of case that judicial reassignment is unnecessary). However, this approach is by far the minority approach. See U nited States v. Rivera, 357 F.3d 290, 297 (3d Cir. 2004) (if specific performance is sufficient remedy, defendant must be resentenced by different judge); Clark, 55 F.3d at 14 (same); United States v. Peglera, 33 F.3d 412, 415 (4th Cir. 1994) (judicial reassignment is “required”); United States v. Goldfaden, 959 F.2d 1324, 1329 (5th Cir. 1992) (defendant “must be sentenced by different judge” if “specific performance is called for”); United States v. Yesil, 991 F.2d 1527, 1533 n.7 (11th Cir. 1992) (following “practice of remanding to a different sentencing judge following the breach of a plea agreement”); United States v. M cCray, 849 F.2d 304, 306 (8th Cir. 1988) (per curiam) (defendant is “entitled” to be resentenced by a different judge); United States v. Brody, 808 F.2d 944, 948 (2d Cir. 1986) (requiring judicial reassignment). Even the Ninth Circuit, which previously subscribed to the minority approach, has recognized (continued...) 25 U.S. at 263; Cachucha, 484 F.3d at 1271. Santobello did not prohibit same-judge resentencing in all instances. Cf. Santobello, 404 U.S. at 263 (noting that “petitioner should be resentenced by a different judge” (emphasis added)). And, in this instance, resentencing by the same judge will in no way threaten the institutional principles that motivated Santobello's call for judicial reassignment, particularly because it is the defendant – the individual victim of the breach – who is requesting this remedy. Indeed, M r. VanDam's representation that “the original judge can be fair” eliminates even the appearance of inequity. Aplt. Br. at 26. Finally, we note that the government has not challenged the same-judge component of M r. V anDam’s requested remedy. III. C ON CLU SIO N For the preceding reasons, we VAC ATE M r. VanDam's sentence and R EM A N D for resentencing before the same sentencing judge. During the course of the resentencing proceeding, the government must specifically perform its obligation to recommend a term of imprisonment at the low end of the Guidelines range that the district court finds to be applicable. 8 (...continued) more recently the controlling nature of Santobello's prescription of judicial reassignment. See Mondragon, 228 F.3d at 981; Camper, 66 F.3d at 232. 26 06-4104, United States v. VanDam L UC ER O, J., concurring in part and dissenting in part. I join all but Parts II.B, II.C, and III of the majority opinion. On the rare occasions in which a defendant does not seek to vacate his guilty plea, but only requests specific performance of the plea agreement, I would conduct harmless error analysis. I disagree with the majority’s conclusion that Santobello v. New York, 404 U.S. 257, 262-63 (1971), controls the outcome of this case. This Circuit has previously recognized a “very limited context” in which plea-agreement breaches are subject to harmless error analysis: when “the complained-of effect of the government’s breach cannot be remedied by resentencing.” U nited States v. Belt, 89 F.3d 710, 713 (10th Cir. 1996). Although the Belt exception is not applicable here, that case forecloses a strict rule barring harmless error review whenever the government breaches a plea agreement. I would create a second limited exception under the circumstances of this case. The purpose of the remand in Santobello was to allow the state court to determine whether voiding the plea agreement or simple resentencing would be appropriate. The full text of the Court’s analysis is as follow s: W e need not reach the question whether the sentencing judge would or would not have been influenced had he known all the details of the negotiations for the plea. He stated that the prosecutor’s recommendation did not influence him and we have no reason to doubt that. Nevertheless, we conclude that the interests of justice and appropriate recognition of the duties of the prosecution in relation to promises made in the negotiation of pleas of guilty will be best served by remanding the case to the state courts for further consideration. The ultimate relief to which petitioner is entitled we leave to the discretion of the state court, which is in a better position to decide whether the circumstances of this case require only that there be specific performance of the agreement on the plea, in which case petitioner should be resentenced by a different judge, or whether, in the view of the state court, the circumstances require granting the relief sought by petitioner, i.e., the opportunity to withdraw his plea of guilty. W e emphasize that this is in no sense to question the fairness of the sentencing judge; the fault here rests on the prosecutor, not on the sentencing judge. 404 U.S. at 262-63. As I read this passage, it has relatively little to do with the propriety of harmless error analysis; instead it primarily concerns the identity of the remedial decisionmaker. Indeed, the preeminent question in the case was who decides whether a defendant may withdraw from a plea agreement following a government breach. Id. at 257-58 (“W e granted certiorari in this case to determine whether the State’s failure to keep a commitment concerning the sentence recommendation on a guilty plea required a new trial.”). Justice M arshall, joined by Justices Brennan and Stewart, argued that the defendant himself should decide. Id. at 268-69 (M arshall, J., dissenting). Those justices would have allowed defendants to choose to void a plea agreement when breached by the government. Justice Douglas concurred in the majority opinion, but wrote separately to counsel that “a court ought to accord a defendant’s preference considerable, if not controlling, weight.” Id. at 267 (D ouglas, J., concurring). A majority of the Court concluded that the state court was “in a better position” to -2- make that determination. Id. at 263. Although the Santobello Court does not explicitly state that harmless error analysis is inapplicable to breach-of-plea-agreement cases, the majority contends that the Court necessarily decided that Santobello was entitled to some relief before addressing what type of relief would be appropriate. I disagree. The Court first concluded that the state tribunal was the appropriate body to decide whether Santobello should be permitted to withdraw his guilty plea. Having no “supervisory jurisdiction” over the state court that would make the withdrawal decision, id. at 266 (Douglas, J., concurring) (quotation omitted), the Court simply vacated the judgment below, id. at 263. Once that judgment was vacated, resentencing (along with specific performance of the plea agreement) would be necessary if the state denied Santobello the opportunity to void his plea. Although Santobello would receive some form of relief from the state court under either route, that result was not based on the Court’s aversion to harmless error review, but the happenstance of procedural posture. I share the majority’s concern over the appearance of institutional integrity, and those countervailing interests make this a difficult issue. Harmless error analysis is rife with the potential for abuse; as my respected colleague recently noted, we must avoid building “an ever-increasing floor under [an] unlawful practice rather than placing a ceiling upon it.” United States v. Nash, 482 F.3d 1209, 1221 (10th Cir. 2007) (M cKay, J., dissenting). Nevertheless, under the -3- unique circumstances present in this case, I w ould conduct harmless error review. -4-