This is a sentencing appeal. Appellant John Baird is a former Philadelphia police officer who was assigned to the infamous 5 Squad of the 39th district. The 5 Squad was responsible for breaking up drug trafficking operations in an area of Philadelphia in which drug dealing has been epidemic. Instead of working to uphold the law, Baird, and those of his police officer colleagues who were also corrupt, systematically broke it. Over the course of a number of years, and in instances too numerous to chronicle here, they executed illegal searches, detained individuals without legal cause, employed excessive force against detainees, caused the false prosecution of numerous individuals, and stole money and property from persons they were investigating. A seven-count indictment details at least forty-five such instances.

As of this writing, the City of Philadelphia is still endeavoring to right all the wrongs caused by the 5 Squad. Victims of 5 Squad corruption have lodged numerous civil suits against the police department and the city, and settlements are costing the city large sums. The District Attorney has been reviewing prosecutions arising from 5 Squad activities. That review has led to the release from prison of a number of innocent persons whose convictions rested on evidence wrongfully obtained or fabricated by 5 Squad officers. A recent newspaper article reported that, as a result of this corruption, the Philadelphia District Attorney has dismissed 160 cases and the city has paid out more than $3.5 million.1

Pursuant to a plea bargain, Baird pled guilty to three counts of the indictment. The calculation of Baird's sentence under the Sentencing Guidelines produced an adjusted offense level of 29 and a criminal history category of I for a sentencing range of 87 to 108 months. The district court might have departed downward from the range pursuant to the government's motion under § 5K1.1 of the Guidelines to reward Baird's substantial assistance to authorities. Baird's cooperation was in fact very great, for he produced evidence against his 5 Squad co-conspirators, leading to numerous arrests. The district court nonetheless effectively denied the § 5K1.1 motion; it also factored Baird's cooperation into the sentence in an unusual manner. Instead of departing downward, the district court departed upward, imposing a sentence of 156 months (13 years), while making it clear that it would have imposed an even greater sentence but for Baird's cooperation. Importantly for this appeal, the court, in fashioning the upward departure, relied upon conduct underlying dismissed counts.

Baird, sorely aggrieved by the perception that his cooperation netted him not a decrease but an increase in his sentence, has appealed on three grounds. First, he contends that the district court erred in considering in connection with the upward departure the conduct underlying counts dismissed as part of a plea agreement. Second, he submits that the upward departure was itself improper. Third, he challenges the extent of the upward departure as unreasonable in light of the treatment of analogous situations under the Sentencing Guidelines.

Although it would seem that after almost ten years of experience under the Guidelines the dismissed counts issue should have been resolved, unfortunately it has not. There exists a circuit split on the issue, and our own jurisprudence, though generally recognizing the appropriateness of using conduct underlying dismissed counts, is clouded by a recent decision suggesting the opposite. Moreover, whatever the general rule, Baird argues that the plea bargain sections of the Guidelines proscribe the use of dismissed conduct to support an upward departure in the case of a bargained plea. We conclude, however, that even in the plea bargain context, conduct underlying dismissed counts may support an upward departure. We have no difficulty with the remaining issues, believing that an upward departure was warranted in this case, and that the extent of the upward departure was not unreasonable. Accordingly, we affirm.

As we have noted, Baird pled guilty to three counts of the indictment. Count Two charged Baird and his codefendants with Hobbs Act robbery, see 18 U.S.C. § 1951, in connection with an illegal search of a house. During the search, the officers seized cash from those present, including from suspected drug dealer Edwin Scott. The officers never reported the seizure. Count Five charged Baird with conspiracy to violate the civil rights of another while acting under the color of state law. See 18 U.S.C. § 241. During this incident, Baird and a codefendant illegally detained Arthur Colbert, threatened him, physically assaulted him, and then conducted an illegal search of his apartment. Count Six charged Baird with obstruction of justice. See 18 U.S.C. § 1503. This count is based on Baird's attempt to mislead investigators into believing that a codefendant was not involved in the 5 Squad corruption.

The terms of the plea agreement made explicit Baird's sentencing exposure. A maximum prison term of 20 years accompanied Count Two; a maximum of 10 years accompanied Count Five; and a maximum of 5 years accompanied Count Six. In all, according to the plea agreement itself, "the total maximum sentence which could be imposed is a term of imprisonment of 35 years...." Moreover, the plea agreement specifically noted that "[n]o one has promised or guaranteed to the defendant what sentence the Court will impose." The government did, however, agree that it would move to allow the sentencing court to depart from the Sentencing Guidelines pursuant to § 5K1.1 if, in its sole discretion, the government determined that Baird fully and truthfully cooperated with the prosecution in its investigation into the activities of the 5 Squad. Because of Baird's extensive cooperation, the government made such a motion.

The computation of Baird's sentence under the Sentencing Guidelines is complicated because it involves the grouping of counts. We need not concern ourselves with the intricacies of this process, however, except to note those specific offense characteristics and adjustments the district court took into account at sentencing. They include the following: a five-level increase under § 2B3.1(b)(2)(C) for the use of a handgun; a four-level increase under § 2H1.1(b)(1) because Baird was a public official at the time of the offense; a two-level increase under § 3A1.3 for restraining the victim of an offense; a two-level increase under § 3B1.1(c) for a supervisory role in the offense; a two-level increase under § 3B1.3 for abuse of a position of trust; a two-level increase under § 3C1.1 for obstruction of justice; and a three-level decrease under § 3E1.1(a), (b) for acceptance of responsibility. The calculation led to a total offense level of 29. Because Baird had no prior criminal history, his criminal history category was I. Under the relevant Sentencing Guidelines, an offense level of 29 with a criminal history category of I yields an adjusted guideline range of 87 to 108 months imprisonment.

In fashioning its sentence, the district court focused heavily on the significant disruption of governmental functions caused by Baird's activities, finding it so great as to justify a departure under § 5K2.7, which provides:

If the defendant's conduct resulted in a significant disruption of a governmental function, the court may increase the sentence above the authorized guideline range to reflect the nature and extent of the disruption and the importance of the governmental function affected.

1995 U.S.S.G. § 5K2.7.

For example, the court observed that "every time a search is undertaken in violation of the Constitution, that is a significant disruption of Governmental function." As further evidence of the disruption, the court cited the many convictions that have been set aside because they were based on illegally obtained evidence, and the civil litigation that has resulted from these wrongful prosecutions. It concluded that the drafters of the Sentencing Guidelines simply did not foresee police corruption of the type and colossal extent involved here. Therefore, it departed upward and sentenced Baird to 156 months for Count Two, 120 months for Count Five, and 60 months for Count Six. The sentences were to run concurrently, so that Baird's total sentence was 156 months, or 13 years.

It will be instructive to recite the relevant portions of the district court's reasoning:

I have considered the guidelines. I do find that there has been a significant disruption of a Governmental function here. I find there to have been many significant disruptions of many Governmental functions. And in my view, every time a search is undertaken in violation of the Constitution, that is a significant disruption of Governmental function.

When somebody chooses to squish the Bill of Rights into the mud, that is a significant disruption of Governmental function, whatever one thinks about the exclusionary rule, and sage people have debated all sides of that issue, our Supreme Court has spoken on that. And when one does that and does that repeatedly, each and every such event constitutes a significant disruption of a Governmental function, whether somebody later tattle-tales on it or not. That it was just a little tree falling in the wilderness, a little unconstitutional search that some police officer decided not to talk about, that disruption occurred then and there.

That because of this defendant's prodigious memory, his willingness to come forth, many of these illicit searches have now been overturned, and many, perhaps some substantively in the Government's sentencing memorandum, suggests all of them substantively ultimately justifiable searches from the standpoint of whether the drugs actually were there, in most of these cases, whether they pan out with continuing convictions or whether those convictions are destroyed, and consequent civil suits, et cetera, that is another aspect of the disruption of governmental function, sort of the second stage of it.

First, is the Constitutional violation and recognition of it, then the disruption of the numerous convictions, then the lawsuits for people whose Constitutional rights were violated.

Governmental functions in this town have been disrupted immeasurably. And I am constrained to conclude that the guidelines simply, as all-encompassing the seemingly all-foreseeing scriveners of those guidelines may have been, didn't fully encompass and embrace the breadth of the criminality before us today.

And I am thus, upon this reasoning, of the view that an upward departure is appropriate. Again perhaps toying with semantics here, I am perhaps in trying to honestly convey to you my thinking on the subject, there is implicit in what I am going to impose something of a downward departure.

I would be hitting you harder, frankly, were it not for the cooperation. I recognize the irony in his zeal to cooperate and cheapen his sentence of being hit hard, but actually the jail, the prisons of the Commonwealth and the United States of America from the Atlantic to the Pacific, are full of people who have been incarcerated for a lot longer than this man is going to go, simply because they got talkative in the Roundhouse [the Philadelphia police administration building] or some similar edifice in some other municipality.

But I think under the circumstances of this case an upward departure is not only appropriate but essential. Upon release from imprisonment, the defendant shall be placed on supervised release for a term of three years.

The sentence actually given to Baird far exceeded his sentencing range as calculated under the Guidelines. In effect, the upward departure was either a four-level increase to an offense level of 33, criminal history category of I, and a sentencing range of 135 to 168 months; or a five-level increase to an offense level of 34, criminal history category of I, and a sentencing range of 151 to 188 months. As is evident from the passages quoted above, the precise anatomy of the actual departure is unclear because the district court was general rather than specific in explaining the departure justification.

Baird timely appealed from the judgment. The district court properly exercised jurisdiction over the matter under 18 U.S.C. § 3231; we exercise appellate jurisdiction over the final judgment of the district court under 18 U.S.C. § 3742 and 28 U.S.C. § 1291.

II. MAY CONDUCT UNDERLYING DISMISSED COUNTS BE CONSIDERED?

Baird's first contention on appeal is that the district court improperly departed upward from the Guideline range based on conduct underlying counts to which Baird did not plead guilty and which were dismissed by reason of the plea agreement.2 As we have noted, the point is still somewhat controversial; four circuits allow consideration of dismissed counts, and two others do not. See infra, part II, section B. Our analysis of this contention, therefore, requires close scrutiny of a number of provisions in the Sentencing Guidelines. We undertake such scrutiny in light of Koon v. United States, --- U.S. ----, 116 S. Ct. 2035, 135 L. Ed. 2d 392 (1996), which teaches that appellate review of a district court's decision to depart from the Guidelines is essentially subject to an abuse of discretion standard. See id. at ---- - ----, 116 S. Ct. at 2046-48. Any legal error in applying the Guidelines would constitute such an abuse. See id. at ---- - ----, 116 S. Ct. at 2047-48.

Whether a sentencing court can consider conduct underlying dismissed counts is at bottom a legal question about which there continues to be not only disagreement but also a certain amount of confusion. As we proceed, we bear in mind that not only is each guideline legally binding on the courts, see Mistretta v. United States, 488 U.S. 361, 391, 109 S. Ct. 647, 664, 102 L. Ed. 2d 714 (1989), but so too are the accompanying policy statements, see Williams v. United States, 503 U.S. 193, 199-201, 112 S. Ct. 1112, 1118-20, 117 L. Ed. 2d 341 (1992), and commentary, see Stinson v. United States, 508 U.S. 36, 41-44, 113 S. Ct. 1913, 1917-18, 123 L. Ed. 2d 598 (1993), so long as neither the policy statements nor the commentary is inconsistent with any guideline.

The Guidelines afford sentencing courts considerable leeway as to the information they may consider when deciding whether to depart from the Guideline range. Section 1B1.4 specifically states that in determining whether a departure is warranted, "the court may consider, without limitation, any information concerning the background, character and conduct of the defendant, unless otherwise prohibited by law." 1994 U.S.S.G. § 1B1.4. Moreover, with respect to conduct underlying dismissed counts, the commentary to § 1B1.4, when read in conjunction with the commentary to § 1B1.3, indicates that considering such conduct is appropriate. The commentary to § 1B1.4 declares that "information that does not enter into the determination of the applicable guideline sentencing range may be considered in determining whether and to what extent to depart from the guidelines." Id. § 1B1.4, comment., backg. And, the commentary to § 1B1.3 states that "[c]onduct that is not formally charged or is not an element of the offense of conviction may enter into the determination of the applicable guideline sentencing range." Id. § 1B1.3, comment., backg.3

It must be, therefore, that conduct not formally charged or not an element of the offense can be considered at sentencing; if such information can be considered in determining the applicable guideline range under § 1B1.3, then such information can be considered in determining whether to depart from that range under § 1B1.4. Thus, conduct underlying dismissed counts--which is conduct that is neither formally charged nor an element of the offense--may be considered at sentencing.4

This conclusion is supported by the weight of the case law. The leading case is United States v. Kim, 896 F.2d 678 (2d Cir.1990) (allowing the consideration of conduct underlying dismissed counts). In Kim, the Second Circuit identified four ways in which the Guidelines addressed misconduct not resulting in conviction. See id. at 682-84. First, the Guidelines take cognizance of acts that typically accompany or occur in the course of an offense. These specific offense characteristics determine the severity of the offense. Second, the Guidelines create rules concerning misconduct to which the parties stipulate in connection with the entry of a plea. Third, calculating the criminal history category under the Guidelines requires an analysis of misconduct not resulting in a conviction. Finally, the Guidelines envision departures based on misconduct not resulting in conviction. Against this background, the court in Kim concluded that "with respect to acts of misconduct not resulting in conviction, the [Sentencing] Commission intended to preclude departures for acts bearing no relationship to the offense of conviction, but to permit departures for acts that relate in some way to the offense of conviction, even though not technically covered by the definition of relevant conduct." Id. at 684; accord United States v. Barber, 93 F.3d 1200, 1209-12 (4th Cir.1996); United States v. Big Medicine, 73 F.3d 994, 997-98 (10th Cir.1995); United States v. Ashburn, 38 F.3d 803, 807-08 (5th Cir.1994) (en banc), cert. denied, --- U.S. ----, 115 S. Ct. 1969, 131 L. Ed. 2d 858 (1995); United States v. Zamarripa, 905 F.2d 337, 341-42 (10th Cir.1990).

In a related vein, the Supreme Court recently held that a sentencing court is permitted to consider conduct of which a jury acquitted a defendant. See United States v. Watts, --- U.S. ----, 117 S. Ct. 633, 136 L. Ed. 2d 554 (1997).5 In so doing, the Court engaged in a close analysis of statutory authority, the Sentencing Guidelines (particularly § 1B1.3 and § 1B1.4), its own precedent, and pre-Guidelines sentencing practices. See id. at ---- - ----, 117 S. Ct. at 635-36. With sweeping language, the Court made plain that a sentencing court is entitled to examine a broad range of factors that may relate to the defendant's conduct, including, but apparently not limited to, the defendant's life, characteristics, and past criminal behavior, even if such behavior did not result in criminal convictions. See id. at ----, 117 S.Ct. at 635. According to the Court, the Guidelines kept in place a sentencing system in which there was no "basis for the courts to invent a blanket prohibition against considering certain types of evidence at sentencing." Id.6

Watts strongly suggests that a sentencing court may consider conduct underlying dismissed counts. If a sentencing court can consider conduct that a jury, after trial, believed to be unproven beyond a reasonable doubt, it would surely seem that a sentencing court can consider conduct underlying a count that, by plea agreement, has been dismissed without having been tested in an adversarial process. We note that in Watts the sentencing courts considered acquitted conduct when calculating the applicable guideline range. Although neither of the underlying cases in Watts was a departure case, we find nothing in Watts that implies that the Supreme Court would treat a departure case any differently. In fact, the Court relies heavily on § 1B1.4, which governs the information applicable to departures, in reaching its conclusion.

The foregoing analysis does not lead ineluctably to the conclusion that any conduct underlying a dismissed count may be considered at sentencing. The conduct underlying the dismissed counts must be related in some way to the offense conduct. To be related conduct, the conduct need not, however, fit the definition of relevant conduct under § 1B1.3.7 Without attempting to define with precision the meaning of "related," we again follow United States v. Kim, supra, and hold that the acts in question must exhibit commonalities of factors sufficient to allow for a reasonable grouping of the separate, individual acts into a larger, descriptive whole. It is not enough, however, that the acts stand in close temporal relation to one another. Rather, the similarities of the acts must arise from the character or type of the acts.

The Second Circuit's logic in Kim in this respect is compelling. It began by noting that specific offense characteristics for a given guideline represent typical occurrences during the commission of the specific crime covered by the guideline. See id. at 682. It went on to note that the relevant conduct guideline, § 1B1.3, is limited to conduct that is somehow related to the offense of conviction. See id. at 682-83. Then, the court noted that in determining the criminal history category a court is directed to similar, prior conduct under § 4A1.3(e). See id. at 683. Finally, the court looked at the language the Guidelines employ with respect to departures and pointed out that departures should only be based on conduct related to the offense of conviction. See id.

The court concluded that these provisions, when taken together, make clear that a sentencing court may only consider dismissed conduct if it is related to the charged conduct. In the case at bar, it is clear that the conduct underlying the dismissed counts was sufficiently related to the charged conduct. All of the acts, whether charged or not, involved essentially the same type of illegal searches and seizures and essentially the same perpetrators. The only differences appear to be the identity of the particular victims.8

The Ninth Circuit in United States v. Castro-Cervantes, 927 F.2d 1079 (9th Cir.1990) reached a conclusion different from the Second Circuit. It held that a court could not consider the conduct underlying dismissed counts--whether related to the offense conduct or not--when making a departure determination. See id. at 1081-82. (A more complete discussion of Castro-Cervantes is more appropriately taken up later. See infra part II, section D.); see also United States v. Harris, 70 F.3d 1001, 1003-04 (8th Cir.1995) (holding that consideration of conduct underlying dismissed counts to support an upward departure is inappropriate). Notwithstanding its Castro-Cervantes holding, the Ninth Circuit also made clear, in United States v. Fine, 975 F.2d 596 (9th Cir.1992), that the rule in Castro-Cervantes only applies to departures based on conduct underlying counts dismissed pursuant to a plea agreement; Castro-Cervantes did not limit consideration of such conduct as relevant conduct in determining the applicable sentencing range. See id. at 602.

C. Does United States v. Thomas Undermine Our Conclusion?

Before proceeding further, we must consider whether our decision in United States v. Thomas, 961 F.2d 1110 (3d Cir.1992), which arguably calls into question our conclusion that consideration at sentencing of conduct underlying dismissed counts is permissible, undermines our conclusions. In Thomas, in the course of obtaining firearms from a gun shop, the defendant falsely stated on four Bureau of Alcohol, Tobacco and Firearms forms that he had never been convicted of a crime punishable by a prison term in excess of one year. In fact, the defendant had a lengthy criminal record. As part of a plea agreement, the defendant agreed to plead guilty to four counts of making false statements in connection with the purchase of a firearm. In exchange, the government agreed not to charge him with possession of a firearm by a convicted felon, a crime that carried with it a mandatory minimum sentence of 15 years. At sentencing, the district court departed from the applicable Sentencing Guideline range of 24-30 months. It justified the departure by its finding that the defendant's criminal history category, the highest category under the Guidelines, under-represented the defendant's criminal history because it did not take into account juvenile convictions, the likelihood of recidivism, and parole revocations.

As an alternative argument supporting the decision to depart, the government argued on appeal that, even assuming that the factors on which the district court relied to depart from the criminal history category were improper, the court could have departed based on the conduct not charged. The Thomas panel disagreed; it wrote that a sentencing court could not depart from the Guidelines to account for conduct underlying a forgone charge. See id. at 1121 ("It would be a dangerous proposition to allow district courts to base upward departures on crimes that were not actually charged.").

Although Thomas seems to suggest that consideration at sentencing of conduct underlying dismissed counts is improper, we believe that any such conclusion is incorrect. We have explained how our conclusion flows from Watts, a Supreme Court decision. To the extent that Thomas is inconsistent with Watts, Thomas will of course be deemed without effect.9 Even assuming that Watts is not controlling because Watts and Thomas arose in different factual contexts--the issue in Watts was the consideration of acquitted conduct and that in Thomas was of dismissed conduct--Thomas is distinguishable from the present case. The particular facts of Thomas and our irritation with the government's position relative to those facts largely determined the result.

Thomas presented an unusual plea agreement. The government "expressly declined to prosecute" the defendant for illegal possession of a firearm. See Thomas, 961 F.2d at 1124 n. 1 (Greenberg, J., concurring). To prosecute would have triggered the provisions of the armed career criminal act. See id. at 1112. The armed career criminal act operates much like an "on/off switch" in that it applies automatically if its prerequisites are met. See id. at 1122. Moreover, once it applies, the armed career criminal act imposes a mandatory minimum sentence; there is no incremental increase in the associated penalties. See id. By arguing for consideration of the conduct underlying the forgone charge, the government in effect reneged on its agreement, and sought to alter the armed career criminal act and impose on it a sliding scale it does not have, ie., change the fundamental nature of the act so that it operated like a dimmer switch. The government did so by arguing that an upward departure--to a sentence less harsh than that mandated by the armed career criminal act--was warranted because it had not sought application of the armed career criminal act. The government wanted the armed career criminal act to play a role in sentencing, but not too much of a role. The court in Thomas plainly disapproved of this attempt.10

Thomas, as we read it, would not foreclose all consideration of conduct underlying dismissed counts. Instead, Thomas seems to be directed at the particular circumstances presented by that case. See id. at 1124 n. 1 (Greenberg, J., concurring) ("I do not understand the opinion to preclude in all circumstances the possibility of a departure predicated on conduct which could have been the basis for additional charges but was not. Here there is a special situation...."). Because the circumstances of the present case differ substantially from those of Thomas, Thomas should not control our inquiry.11

D. Import of the Policy Statement Concerning Plea Agreements

Baird contends alternatively that the policy statement outlining the standards for accepting a plea agreement prohibits, as a matter of law, consideration of conduct underlying a dismissed charge. The relevant portion of the policy statement follows:

In the case of a plea agreement that includes the dismissal of any charges or an agreement not to pursue potential charges [Rule 11(e)(1)(A) ], the court may accept the agreement if the court determines, for reasons stated on the record, that the remaining charges adequately reflect the seriousness of the actual offense behavior and that accepting the agreement will not undermine the statutory purposes of sentencing or the sentencing guidelines.

Provided, that a plea agreement that includes the dismissal of a charge or a plea agreement not to pursue a potential charge shall not preclude the conduct underlying such charge from being considered under the provisions of § 1B1.3 (Relevant Conduct) in connection with the count(s) of which the defendant is convicted.

1994 U.S.S.G. § 6B1.2(a).

According to Baird, the first paragraph of this subpart means that, if the court believed that the counts not dismissed by virtue of the plea agreement did not satisfactorily account for the seriousness of the actual offense, then what the court should have done is to reject the plea agreement. In other words, Baird asserts, conduct underlying dismissed counts provides grounds for rejecting a plea agreement but not for departing from the applicable sentencing range.

Baird draws support for this argument from Castro-Cervantes. The Ninth Circuit in Castro-Cervantes reasoned that if a plea agreement does not reflect the seriousness of the offense, the court should reject the agreement at the outset; at sentencing, it is too late to seek to address shortcomings in the plea agreement. Allowing a court to depart from the sentencing range based on conduct underlying dismissed counts would "violate[ ] the spirit if not the letter of the bargain" represented by the plea agreement. Castro-Cervantes, 927 F.2d at 1082. The Ninth Circuit expanded on the policy justifications driving Castro-Cervantes by noting that "allow[ing] judges to depart from the Guidelines on the basis of counts that have been dropped pursuant to plea agreements would severely undermine the incentive of defendants to enter into plea agreements." United States v. Faulkner, 952 F.2d 1066, 1070 (9th Cir.1991).12

We disagree. This argument begs the ultimate question. Whether a court accepts a plea agreement is dependent, to some extent, on the information it can consider at sentencing. If the court is aware that it cannot consider conduct underlying dismissed counts at sentencing, then it may be more reluctant to accept a plea agreement. The opposite is also true. If the court is aware that it may consider conduct underlying dismissed counts at sentencing, then it may more readily accept a plea agreement. At all events, we do not see Baird's argument from § 6B1.2(a) as undercutting the conclusion we have already reached regarding the use of dismissed conduct. The language of the policy statement is insufficiently specific to contradict the clearer guidance provided by other sections of the Guidelines and the Supreme Court, discussed supra.

Baird also argues that the second paragraph of this policy statement implies that conduct underlying dismissed counts may only be used to determine the applicable sentencing range under § 1B1.3, but may not be used to determine whether to depart from that range. Otherwise, the argument continues, the first paragraph of this subpart would be mere surplusage; no court would hesitate to accept a plea agreement that includes dismissed counts if that court could always consider the conduct underlying those dismissed counts.

We disagree. As noted above, § 1B1.3, § 1B1.4, and the commentary thereto make clear that the information appropriately considered in a decision to depart is broader than that considered in determining the applicable guideline range. If § 6B1.2(a) does not preclude a court from examining conduct underlying dismissed counts in determining the applicable sentencing range (in fact, the proviso contained in the last paragraph of § 6B1.2(a) seems to encourage it), then we see no reason why it should preclude the court from examining the same information when deciding whether to depart.13

Moreover, even if a court actually considers conduct underlying a dismissed count, it does not automatically ensure that the remaining charges will reflect the seriousness of the actual offense conduct. The statutory maximum sentence for the remaining charges may be relatively low; in such a case, it might be of little significance that a court could consider conduct underlying dismissed counts because that court would be unable to impose a sentence in excess of the statutory maximum. Under our interpretation of the Guidelines, then, a court might reject a plea agreement because it believes that the statutory maximum sentence for the remaining counts is too short to account both for the charged conduct and for the dismissed conduct. Contrary to Baird's submission, then, the first paragraph of § 6B1.2(a) is not mere surplusage.

Our conclusion that conduct underlying dismissed counts may be considered when determining whether to depart from the applicable Sentencing Guidelines range comports with the policies underlying the Guidelines themselves. The Guidelines are, at bottom, a modified real offense system. See 1994 U.S.S.G. chap. 1, pt. A, intro. comment. 4(a). More specifically, they are a mix of a charge offense system and a pure real offense system in that it bases a sentence on both the formal offense of conviction and on the actual conduct of the defendant. See Stephen Breyer, The Federal Sentencing Guidelines and the Key Compromises Upon Which They Rest, 17 Hofstra L.Rev. 1, 8-12 (1988). Therefore, it is clear that the Guidelines envisioned that sentencing courts would consider at least some conduct for which a defendant was not actually charged.

We are unconvinced by Baird's argument that use of conduct underlying dismissed counts will deny defendants the benefit of the plea agreement bargain, nor, as we explained, see supra part II, section D, do we find persuasive the contention that without that benefit there will be no incentive for defendants to plea bargain. In the usual case, the plea agreement makes clear that the sentencing court is not bound by the agreement. Within statutory bounds, the sentencing court has great discretion. If a defendant is sentenced to a term of imprisonment within the maximum set out in the plea agreement, it is difficult to see the grounds on which a defendant can rest a complaint; the defendant got what he bargained for.14

At all events, the incentive to plea bargain remains. The defendant can limit the sentencing court's discretion by bargaining to plead guilty only to charges with lower statutory maxima. Of course, the court may reject that plea agreement. And, if the defendant pled pursuant to Federal Rule of Criminal Procedure 11(e)(1)(A)--which allows the government to dismiss charges--or 11(e)(1)(C)--which allows the government and defendant to agree on a specific sentence--he can withdraw his plea.15 Moreover, the Sentencing Guidelines allow for an adjustment of offense level for the acceptance of responsibility, see 1994 U.S.S.G. § 3E1.1, and courts routinely make this adjustment for defendants who plead guilty.

We conclude that a sentencing court, in its determination whether to depart from the sentencing range made applicable by the Sentencing Guidelines, may consider conduct underlying counts dismissed pursuant to a plea agreement, provided that such conduct is related to the conduct forming the basis of the remaining counts and that such conduct is proved by at least a preponderance of the evidence. In this case, it was appropriate for the district court to consider conduct underlying the counts against Baird that were dismissed. That conduct was related to the charged conduct and it was proved by at least a preponderance of the evidence.

III. DID THE DISTRICT COURT ABUSE ITS DISCRETION IN DECIDING

TO DEPART?

We now must determine whether, given the information before it, the district court abused its discretion by departing from the applicable guideline range. Unlike our discussion heretofore, the legal standard is relatively clear, and our focus will be on the particular factual circumstances presented by this case.

Section 5K2.0, Grounds for Departure, provides a roadmap for a decision to depart from the applicable Guideline range. In the usual case, a sentencing court must impose a sentence within the Guideline range. See 18 U.S.C. § 3553(b). For the most part, a court can "treat each guideline as carving out a 'heartland,' a set of typical cases embodying the conduct that each guideline describes." See 1994 U.S.S.G. ch. 1, pt. A, intro. comment. 4(b). However, the Sentencing Commission recognizes that "it is difficult to prescribe a single set of guidelines that encompasses the vast range of human conduct potentially relevant to a sentencing decision." Id. Therefore, a court may depart from the range if it "finds that there exists an aggravating or mitigating circumstance of a kind, or to a degree, not adequately taken into consideration ... in formulating the guidelines." 18 U.S.C. § 3553(b); see 1995 U.S.S.G. § 5K2.0. It is only in the unusual case, one outside the "heartland," in which a departure is authorized. See Koon, --- U.S. at ----, 116 S. Ct. at 2044. Except for a limited number of prohibited factors, the Guidelines do not "limit the kinds of factors, whether or not mentioned anywhere else in the guidelines, that could constitute grounds for departure in an unusual case." 1995 U.S.S.G. ch. 1, pt. A, intro. comment. 4(b).16

"Sentencing courts are not left adrift, however." Koon, --- U.S. at ----, 116 S. Ct. at 2045. The Guidelines list factors that are encouraged as bases for departure and those that are discouraged. Section 5K2.0 introduces the subject. As § 5K2.0 explains, the Commission "seeks to aid the [sentencing] court by identifying some of the factors that [it] has not been able to take into account fully in formulating the guidelines." 1995 U.S.S.G. § 5K2.0. Those factors are included in §§ 5K2.1-5K2.18, and cover issues such as the death or injury of a victim, the harm to property or government interests, the purpose or cause of the offense, the particular conduct of the defendant, and the like. In Chapter 5, Part H, the Guidelines list a number of discouraged factors. These are "factors [that] are not ordinarily relevant to the determination of whether a sentence should be outside the applicable guideline range." 1995 U.S.S.G. ch. 5, pt. H, intro. comment. They include, but are not limited to, a defendant's age (§ 5H1.1), education or vocational skills (§ 5H1.2), or employment history (§ 5H1.5).

In Koon, the Supreme Court explained how a sentencing court is to factor prohibited, encouraged, and discouraged factors into the sentencing decision. If the unusual or special factors presented by a case are prohibited factors, a court may not depart on that basis alone. See Koon, --- U.S. at ----, 116 S. Ct. at 2045. If the unusual or special factors are encouraged factors, a court is merely authorized to depart. See id. A departure employing encouraged factors is commonly referred to as a "guided" departure. If the unusual or special factors are discouraged, or if they are encouraged factors that have been taken into account by the relevant guideline, then a court may depart only if the "factor[§ are] present to an exceptional degree or in some other way make[ ] the case different from the ordinary case where the factor[§ are] present." Id. Finally, a sentencing court may depart even if the unusual or special factors have not been mentioned in the Guidelines, but only in rare circumstances. See id. Such a departure would be "unguided."

B. Can the Disruption of Government Function Serve as a

Basis for Departure in this Case?

Disruption of governmental function is included among the encouraged upward departure factors. See 1994 U.S.S.G. § 5K2.7. "If the defendant's conduct resulted in a significant disruption of governmental function, the court may increase the sentence above the authorized guideline range to reflect the nature and extent of the disruption and the importance of the governmental function affected." Id. Although an encouraged factor, the disruption in this case was, at least in part, arguably taken into account when determining the applicable sentencing range. For example, as part of the calculation of his total offense level, Baird received a two-level increase because, as a police officer, he abused his position of trust. See id § 3B1.3. He also received a four-level increase because, again as a police officer, he was a public official at the time of the offenses. See id. § 2H1.1(b)(1). In other words, Baird's sentence reflected the fact that he was a rogue government official, one who affected government operations by acting outside of the law. Therefore, as stated in Koon, the sentencing court could have departed in this case only if the disruption was "present to an exceptional degree or in some other way [made] the case different from the ordinary case where the [disruption] is present."

We believe that it is clear that the sentencing court did not abuse its discretion by implicitly concluding that the disruption of governmental function caused by Baird's conduct is "present to an exceptional degree." The full impact of Baird's conduct is not yet known. However, based upon information supplied by the City of Philadelphia and by Baird himself, we know that the city has reopened innumerable criminal cases, originating from the 39th District, to determine whether it obtained convictions based on illegally obtained evidence.

The city has already set aside more than one hundred and fifty such convictions, leading to the release of innocent persons from prison. As a result of these wrongful convictions, many individuals have instituted civil lawsuits seeking damages from the city. The city stands to be liable for enormous sums of money. In other words, the disruption Baird caused is not only by no means ordinary, but also is, as far as we can tell, as colossal as it is unprecedented. The city must invest considerable time, vast energy, and enormous resources in making right the wrongs Baird has caused. Therefore, the departure in this case to reflect a disruption in government functions was appropriate.

Baird submits that the disruption occurred as a result of his truthful cooperation with the investigation into the 39th District and not as a result of his unlawful conduct. Therefore, the argument continues, he should not be subject to a departure because § 5K2.7 only allows departures "[i]f the defendant's conduct resulted in a significant disruption...." Id. § 5K2.7 (emphasis added). In a sense, Baird is correct. Without his cooperation, the full extent of the 39th District corruption might never have come to light. We are not unsympathetic, but there is a flaw in Baird's argument: his conduct in fact caused the disruption. But for Baird's illegal activities, the city would not have reopened criminal convictions originating in the 39th District and innocent people would have no cause to sue the city.

In other words, if Baird had been an honest police officer, the city would not need to invest considerable time, energy, or resources in making anything right because nothing was wrong. At all events, we have made clear that the literal language of an encouraged departure factor is not controlling. See Kikumura, 918 F.2d at 1116 ("[F]itting a case within the literal language of a § 5K2 provision is neither a necessary nor a sufficient condition for making an offense-related departure."). That the disruption in this case might have been caused by Baird's cooperation as opposed to Baird's conduct is therefore of little consequence. We hold that a departure in this case for a significant disruption of government functions was not an abuse of discretion.

IV. WAS THE EXTENT OF DEPARTURE APPROPRIATE?

Having concluded that a departure in his case was appropriate, the final stage in our analysis is to determine whether the extent of the departure was reasonable. Our review of the sentencing court's decision in this regard is deferential. See Kikumura, 918 F.2d at 1110; cf. Koon, --- U.S. at ---- - ----, 116 S. Ct. at 2046-48 (holding that a decision whether to depart is reviewed for an abuse of discretion). However, there are "objective standards to guide the determination of reasonableness." Kikumura, 918 F.2d at 1110. Those objective standards can be found in the Guidelines themselves, which provide analogies to which sentencing courts must look when making their determinations. See id. at 1110-14.

In the present case, the district court failed to undertake the analogic reasoning that Kikumura often requires. However, as in Kikumura, our examination of the record leads us to conclude beyond any doubt that even if we were to remand the district court would impose as high a sentence as possible up to 13 years. "If a reasonable analogy existed to support the sentence imposed, remand would be a pointless exercise. We therefore proceed to consider whether such analogy exists." Id. at 1114 (footnote omitted). In so doing, we must bear in mind that "[w]e are dealing here with analogies to the guidelines, which are necessarily more open-textured than applications of the guidelines." Id. at 1113 (emphasis in original).

Our task, then, is to determine if a reasonable analogy exists in the Guidelines that would justify a four- or five-level upward departure based on the disruption of governmental functions. We note the existence of a guideline for the conflict of interest crimes. See 1994 U.S.S.G. § 2C1.3. Under § 2C1.3, the base offense level for the criminal, financial and nonfinancial conflict of interest by federal officials is 6. See id. § 2C1.3(a). However, "[i]f the offense involved actual or planned harm to the government," the Guidelines require an enhancement of 4 levels. See id. § 2C1.3(b)(1). Section 2C1.3 implies, then, that the Guidelines consider harm to the government worthy of a four-level increase. Since Baird's conduct in this case is infinitely worse than a mere conflict of interest that results in harm to the government, a fortiori, the four-level departure in this case was reasonable.17

The judgment of the district court will be affirmed.18