Opinion for the Court filed by Circuit Judge WILLIAMS.
Opinion by Circuit Judge GARLAND concurring in part and dissenting in part.
WILLIAMS, Circuit Judge:
A jury in district court convicted Sonni Wilson of bank fraud and other related offenses. The evidence at trial revealed two separate schemes — one in 1996 targeting several banking institutions including the First Bank Card Center and one in 1997/98 targeting First National Bank of Maryland. In both, Wilson fraudulently procured and used credit cards, ATM cards, check cards and checks issued in other people’s names. In some cases he accomplished his fraud by opening entirely new accounts, while in others he supplied confidential personal information about actual account holders to fraudulently gain control of their accounts. For any one account, the fraud was necessarily short-lived: Use of an actual account would quickly trigger reaction either by the true holder or by bank personnel on the alert [43]*43for suspicious activity such as unusually large cash withdrawals; use of fictional accounts would be exposed by suspicious account activity or by non-payment of the bill.
Wilson was first arrested in 1996 after bank investigators alerted the police. After indictment, he jumped bail. Following a new arrest in 1998, he was charged with six counts of bank fraud (18 U.S.C. § 1344), one count of possession of 15 or more unauthorized access devices1 with intent to defraud (18 U.S.C. § 1029(a)(3)), one count of conspiracy to commit bank fraud and to possess 15 or more unauthorized access devices with intent to defraud (18 U.S.C. § 371), and one count of possession of five or more false identification documents with intent to use illegally (18 U.S.C. § 1028(a)(3)). The jury convicted Wilson on all counts, and the district court sentenced him to 51 months’ imprisonment followed by three years of supervised release.
On appeal Wilson challenges several aspects of his conviction and sentencing. Because of an error in sentencing, we reverse.
;■{ sji :fc ‡ iji 4:
Effect on commerce of Wilson’s possession of access devices. 18 U.S.C. § 1029(a), which prohibits various forms of access device fraud, applies only “if the offense affects interstate or foreign commerce.” Wilson first argues that under United States v. Lopez, 514 U.S. 549, 115 S.Ct. 1624, 131 L.Ed.2d 626 (1995), he can be held criminally accountable under federal law only if the government proves that his actions had a “substantial” effect on interstate commerce. But we have already held, since Lopez, that to support a statutory jurisdictional link for a specific criminal act it is enough that the evidence show that the act had “an ‘explicit’ and ‘concrete’ effect on interstate commerce, rather than a ‘substantial’ one.” United States v. Harrington, 108 F.3d 1460, 1465 (D.C.Cir.1997). In Harrington, we upheld a conviction on the basis of evidence that defendant’s robbery of a Roy Rogers restaurant deprived the restaurant of money that would have traveled to an out-of-state bank and then been used by the Roy Rogers parent company in part to make out-of-state purchases. See id. at 1468.
Wilson also offers a second, independent argument that the evidence failed to show that the access card offenses had any effect at all on interstate commerce. The government concedes that Wilson properly preserved this argument by making a motion for judgment of acquittal after the government rested. Because Wilson presented no defense at all, his motion at the end of the government’s case fully preserved his claim. See United States v. Foster, 783 F.2d 1082, 1085 (D.C.Cir.1986); see also United States v. Sherod, 960 F.2d 1075, 1077 (D.C.Cir.1992). We review de novo the denial of the motion to determine whether the evidence, considered in the light most favorable to the government, was “sufficient to permit a rational trier of fact to find all of the essential elements of the crime beyond a reasonable doubt.” Harrington, 108 F.3d at 1464.
Wilson’s claim is meritless. His only argument for insufficiency is that an expert witness’s general testimony regarding the losses suffered by banks as a result of similar fraudulent schemes was not specific enough to establish that Wilson’s deeds affected interstate commerce. But the government points to a great deal of other evidence that speaks to the interstate commercial effect of Wilson’s fraud. There is, in fact, evidence of interstate impact for all 16 of the devices charged in the indictment.
Of these devices, three are armed services MasterCards that Wilson applied for in 1996 through Andrews Air Force Base [44]*44in Maryland, causing them to be issued by the First Card Bank Center in Louisiana, and sent to Washington, D.C. Regarding the 1997/98 scheme, 12 access devices serviced accounts that Wilson fraudulently opened via phone calls to a bank center in North Carolina but were handled by a Washington, D.C. branch office. (For several of these Wilson used a Maryland address, and for several of those with a Maryland address he used identities of persons located in states other than the District or Maryland). Finally, the remaining device is a Sears Card issued in the name of a California resident and found in Wilson’s wallet when he was arrested in Washington, D.C. Although Wilson evidently lived in Washington, he used this card in Maryland and provided a Maryland address for the account. The jury could reasonably find the modest interstate effect required under Harrington.
Failure to instruct jury on need to find interstate nexus for false ID conviction. Conviction under 18 U.S.C. § 1028 for possession of five or more false identification documents, with intent to use unlawfully, requires that the government satisfy the requirement of § 1028(c)(1) that the possession be “in” or “affect” interstate commerce. The government concedes that the trial judge failed to instruct the jury on the need for such a finding. And Wilson concedes that because of his failure to object at trial the error would be grounds for reversal only if it amounted to plain error under Rule 52(b) of the Federal Rules of Criminal Procedure.2 As the Supreme Court explained in United States v. Olano, 507 U.S. 725, 113 S.Ct. 1770, 123 L.Ed.2d 508 (1993), Rule 52(b) requires that there “be an error that is plain and that affect[s] substantial rights. Moreover, Rule 52(b) leaves the decision to correct the forfeited error within the sound discretion of the court of appeals, and the court should not exercise that discretion unless the error seriously affect[s] the fairness, integrity, or public reputation of judicial proceedings.” Id. at 732, 113 S.Ct. 1770 (internal quotations omitted).
The Supreme Court’s holding in United States v. Gaudin, 515 U.S. 506, 115 S.Ct. 2310, 132 L.Ed.2d 444 (1995), strongly suggests there was an error and that it was “plain.” The Court held that the Fifth and Sixth Amendments require that all elements of a crime be submitted to the jury, with the only conceivable exception being for issues involving a “uniform pos-tratification practice” to the contrary. Id. at 519, 115 S.Ct. 2310. The government invokes no such practice. But we need not conclusively determine the issue, as Wilson has failed to show that the alleged error affected “substantial rights.”
In his opening brief, Wilson argued summarily that the defective instruction affected substantial rights and was “prejudicial”: “It invited the jury to convict without finding whether Appellant’s conduct had been in or affected interstate commerce. There is no reason to believe that the jury disregarded that invitation.” Appellant’s Main Brief at 29. This simple treatment would have been adequate if omission of an essential element of the crime were a “structural” error, such as “complete deprivation of counsel or trial before a biased judge,” which is automatically deemed to affect substantial rights. Neder v. United States, 527 U.S. 1, 8-9, 119 S.Ct. 1827, 144 L.Ed.2d 35 (1999). But Neder holds that failure to instruct on an element of the crime is not structural. Where, as in Ned-er, objection has been made, omission of an element of the crime from the instructions is reviewed for harmless error, id. at 8-15, 119 S.Ct. 1827, so that (the error being of constitutional magnitude) the verdict can be upheld if the government shows “‘beyond a reasonable doubt that the error complained of did not contribute to the verdict obtained.’ ” Id. at 15, 119 S.Ct. [45]*451827 (quoting Chapman v. California, 386 U.S. 18, 24, 87 S.Ct. 824, 17 L.Ed.2d 705 (1967)). Olano explains that for plain error review the burden on prejudice is reversed, requiring the defendant to show the error’s likely effect on the verdict. 507 U.S. at 734, 113 S.Ct. 1770. Wilson made no effort whatever to carry that burden. Thus the government was on solid ground in reading his brief as claiming only structural error. As the alleged error was not structural, Wilson has failed to offer support for a key ingredient of his claim, which thus necessarily fails.
We are not diverted from this conclusion by the fact that the government’s brief, in a backup passage addressing the final element of plain error (whether the error affected “the fairness, integrity or public reputation of judicial proceedings,” see Olano, 507 U.S. at 732, 113 S.Ct. 1770), included a summary collection of evidence on interstate impact. First, as this was offered for a purpose different from the issue of actual impact, the government could fairly suppose that different standards were applicable. Second, we are doubtful in any event whether gilding the lily in the appellee’s brief should ever excuse an appellant’s complete failure to support a necessary ingredient of a claim. Similarly, of course, Wilson’s effort in his reply brief to meet his burden of showing prejudice comes too late. McBride v. Merrell Dow and Pharmaceuticals, Inc., 800 F.2d 1208, 1210 (D.C.Cir.1986); see also Carducci v. Regan, 714 F.2d 171, 177 (D.C.Cir.1983).
Alleged prosecutorial “vouching” for a witness’s credibility. Wilson argues that his conviction should be overturned because the prosecutor, during closing argument, improperly vouched for the credibility of an Inspector Bartlett, who had investigated and arrested Wilson in both 1996 and 1998 and who obtained a confession from him after the 1998 arrest. Although Wilson does not actually identify any specific statement, the target of his complaint appears to be the second half of the prosecutor’s claim that “there is no evidence to support any of those allegations [against Bartlett] and, if it [sic] was, we would not be here.” Trial Transcript (June 15, 1998) at 70. The context of the statement was the defense attorney’s suggestion, as the climax of his closing argument, that because of racial bias Bartlett had not only manufactured the confession but somehow tainted (or perhaps even manufactured) “[t]his case, and all of this evidence.” Id.
We need not decide whether, given this provocation, there was any error in the trial court’s failure to act on the prosecutor’s response. Wilson has shown no impact on “substantial rights,” which Olano requires and here means a demonstration of prejudice. To judge the prejudicial effect of a closing argument error we look to the severity of the alleged misconduct, the centrality of the issue affected by the error, the steps taken to mitigate the error, and the closeness of the case. See United States v. Gartmon, 146 F.3d 1015, 1026 (D.C.Cir.1998). Wilson makes no effort to show how the prosecution’s single offhand remark could have been a severe error. Cf. id. at 1026 (“ Without other compelling factors, a single misstatement confined to a closing argument rarely amounts to severe misconduct.’ ”). While Bartlett’s account of Wilson’s statement and Bartlett’s role in authentication of the fraudulent credit cards and IDs made him an important witness, Wilson’s guilt was critically proven by the documents themselves, and the testimony of several victims, bank fraud investigators, and co-conspirator Clarence Terrell. Wilson concedes that the court gave the standard limiting instructions that the lawyers’ arguments are not evidence, and we have found that such instructions “mitigate the impact of'erroneous jury argument.” Id. at 1026. Finally, the case was not particularly close. We find no prejudice, and thus no plain error.
Effectiveness of counsel. In a pro se brief Wilson asserts that trial coun[46]*46sel was ineffective. We do not normally resolve such claims when raised initially on appeal, unless “the record is so clear that remand is unnecessary.” United States v. Soto, 132 F.3d 56, 59 (D.C.Cir.1997). As Wilson’s claims are vague and conclusory, or based on assertions of admissions by counsel that are plainly not admissions of any ineffectiveness, they do not meet that standard. Pursuant to our usual practice, we remand the issue for consideration by the trial court.
Enhancement of sentence for obstruction of justice. Wilson says that the court erred by imposing a two-level enhancement for obstruction of justice, under U.S.S.G. § 3C1.1, based upon a finding that Wilson committed perjury at a suppression hearing. The court found two separate perjuries. Either would be sufficient, so we need resolve Wilson’s (failing) attack on only one.
At the suppression hearing Wilson falsely denied that he was the man depicted in a photograph shown to him at the hearing. He says that the question was not material to the subject of the hearing, which focused on the voluntariness of Wilson’s post-arrest confession. Materiality is indeed essential, see United States v. Dunnigan, 507 U.S. 87, 94, 113 S.Ct. 1111, 122 L.Ed.2d 445 (1993), but Wilson’s claim is specious. The government presented evidence that at his arrest Wilson claimed to be “Imione Wilson,” that he signed a waiver of his Miranda rights as “Imione Wilson,” and that a driver’s license bearing that name was found in his residence. Wilson denied that he had signed such a waiver and that he was Imione Wilson. When presented with the Imione Wilson driver’s license, he denied that he was the person depicted in the photo on the card. His denial was clearly material to whether or not he had signed the waiver, and thus to whether or not his confession had been voluntary.
Sentence enhancement under U.S.S.G. § SBl.l(a). - Section 3Bl.l(a) of the United States Sentencing Guidelines allows for a four-level upward adjustment in the base offense level “[i]f the defendant was an organizer or leader of a criminal activity that involved five or more participants or was otherwise extensive.” Wilson argues that the trial court erred both when it determined he was an “organizer or leader,” and also when it found the relevant criminal activity was “otherwise extensive.”
When reviewing the district court’s application of the Guidelines, “purely legal questions are reviewed de novo; factual findings are to be affirmed unless ‘clearly erroneous’; and we are to give ‘due deference’ to the district court’s application of the guidelines to facts.” United States v. Kim, 23 F.3d 513, 517 (D.C.Cir.1994); see 18 U.S.C. § 3742(e).
In finding that Wilson was an “organizer or leader of a criminal activity,” the district court relied heavily on the testimony of Clarence Terrell, a former bank teller at First National. Terrell testified that he assisted Wilson’s fraudulent scheme by providing him with names and confidential information regarding account holders. The court found that Wilson “solicited Mr. Terrell’s involvement” in the criminal conduct, and gave him “very explicit directions as to exactly the kind of information, and exactly the kind of profile that he wanted Mr. Terrell to get out of the bank’s records.” Sentencing Transcript (June 10, 1999) at 56. In addition, the court found that “Mr. Terrell himself received extremely little gain from the entire scheme, and Mr. Terrell had no decision making role or authority in the criminal activity.” Id.
Wilson attacks these findings, mainly on the ground that they do not address his “control” of Terrell. We need not attempt an exegesis of the concept of control. Cf. United States v. Kelley, 36 F.3d 1118, 1129 (D.C.Cir.1994). The exercise of decision-making authority, recruitment, and a claimed right to a larger share of the proceeds are prominent among the factors [47]*47that the commentary to the Guidelines indicates should be considered. See U.S.S.G. § 3B1.1, Application Note 4. Given that the trial court’s findings accurately reflect Terrell’s trial testimony, the court’s determination easily meets the “due deference” standard.
Although the finding that Wilson was an organizer or leader of criminal activity is itself enough to justify a two-point enhancement under § 3Bl.l(c), Wilson’s four.-point enhancement under § 3Bl.l(a) is contingent on the additional finding that the criminal activity Wilson organized or led involved “five or more participants, or was otherwise extensive.” “Participants,” for these purposes, explicitly include only persons “criminally responsible for the commission of the offense.” § 3B1.1, Application Note 1. At sentencing, the court conceded that “there may be some question in the evidence as to whether five or more participants were actually clearly established,” but found the criminal activity “otherwise extensive,” declaring that “there is no question in the court’s mind on the basis of ... evidence presented at trial ... that this was a many layered scheme that was an extraordinarily extensive scheme to defraud people of their monies.” Sentencing Transcript (June 10,1999) at 55-56.
Wilson asserts that the district court’s concept of “otherwise extensive” was incorrect, and that to make such a finding the court must look primarily or solely to the number of persons involved in the criminal activity, criminally or noncriminally, as did the Second Circuit in United States v. Carrozzella, 105 F.3d 796, 802 (2d Cir.1997). We agree.
The circuits are currently split on the factors relevant to an activity’s being “otherwise extensive.” The Third Circuit has recently adopted the Carrozzella test. See United States v. Helbling, 209 F.3d 226, 244-45 (3rd Cir.2000). On the other side the leading case is United States v. Dietz, 950 F.2d 50, 53 (1st Cir.1991), which reads the Guidelines as imposing first an “irreducible minimum” requirement that the defendant be involved in criminal activity with at least one other criminally responsible person,3 but, once this is met, as directing the court to plunge into an unconstrained inquiry into the scale of the activity. Thus, under Dietz, the court looks to “the totality of the circumstances, including not only the number of participants but also the width, breadth, scope, complexity, and duration of the scheme.” Id. The Tenth Circuit has expressly endorsed the Dietz test, see United States v. Yarnell, 129 F.3d 1127, 1139 (10th Cir.1997), and others have similarly looked to a broad range of factors beyond the number of persons involved, see, e.g., United States v. Tai, 41 F.3d 1170, 1174-75 (7th Cir.1994); United States v. Rose, 20 F.3d 367, 374 (9th Cir.1994); United States v. Mergerson, 4 F.3d 337, 348 (5th Cir.1993).
We think the Second and Third Circuits have the better case. It is true that the text of § 3Bl.l(a) says nothing about what factors render criminal activity “extensive.” But the Sentencing Commission’s Commentary focuses solely on the role of unkno^ng actors: “In assessing whether an organization is ‘otherwise extensive,’ all persons involved during the course of the entire offense are to be considered. Thus, a fraud that involved only three participants but used the unknowing services of many outsiders could be considered extensive.” U.S.S.G. § 3B1.1, Application Note 3.
Our dissenting colleague correctly notes that the commentary does not expressly state that the number of persons is the “only” relevant factor, see Dissent at 53, but then relies on other language in the commentary to support a broader inter[48]*48pretation of the phrase “otherwise extensive”:
In relatively small criminal enterprises that are not otherwise to be considered as extensive in scope or in planning or preparation, the distinction between organization and leadership, and that of management or supervision, is of less significance than in larger enterprises that tend to have clearly delineated divisions of responsibility. This is reflected in the inclusiveness of § 3Bl.l(c).
U.S.S.G. § 3B1.1, Background (emphasis . added); see Dissent at 53.
While we agree that the reference to “scope,” “planning,” and “preparation” is somewhat confusing, in context the sentence ultimately supports our reading of § 3B1.1. The immediately preceding paragraph of the commentary explains that § 3B1.1 “provides a range of adjustments to increase the offense level based upon [1] the size of a criminal organization (ie., the number of participants in the offense) and [2] the degree to which the defendant was responsible for the offense.” Id. (emphasis and numbers added). The language at issue here, whether a criminal activity “involved five or more participants or was otherwise extensive,” § 3Bl.l(a),(b), plainly relates only to the first factor — “size.” Although the commentary does not explicitly discuss the “otherwise extensive” prong, the explicit identification of “size” with “number of participants” reinforces the impression that the language is concerned with the number of people involved in the offense. On the other hand, § 3B1.1 addresses the second factor, degree of responsibility, by providing enhancement only for the criminal who is either an “organizer or leader” (subsection (a)) or “a manager or supervisor” (subsection (b)).
The paragraph emphasized by the dissent addresses a particular aspect of how § 3B1.1 treats these two basic factors. Subsections (a) and (b) cover the cases where the activity “involved five or more participants or was otherwise extensive,” giving the “organizer or leader” a four-level enhancement, and the “manager or supervisor” only a three-level one. But when the size factor is not satisfied, § 3Bl.l(c) applies a uniform two-level enhancement to the “organizer, leader, manager, or supervisor,” drawing no distinction between these types of responsibility. The language quoted explains the interaction by noting that “the distinction between organization and leadership, and that of management or supervision” is of “less significance” in the case of “relatively small criminal enterprises that are not otherwise to be considered as extensive in scope or in planning or preparation.” § 3B1.1, Background (emphasis added). By contrast, the distinction is more significant in the ease of “larger organizations that tend to have clearly delineated divisions of responsibility.” Id. (emphasis added).
Thus, rather than suggest that scope, planning and preparation actually define an organization as large or small, the commentary merely conveys the point that organizations that are larger tend to be of broader scope and involve more planning and preparation than those that are smaller, and that, for this reason, it is appropriate to distinguish between types of responsibility in larger organizations to an extent not necessary in the case of smaller ones. Far from defining whether or not a criminal activity is “otherwise extensive,” the cited factors are relevant to decide the entirely separate question of degree of responsibility. As the commentary states elsewhere, “[i]n distinguishing a leadership and organizational role from one of mere management or supervision” the court should consider factors such as “the degree of participation in planning or organizing the offense” and “the nature and scope of the illegal activity.” Id. at Application Note 4.
The court in Carrozzella also reasoned that an open-ended approach invited double counting:
[49]*49Many characteristics that might ordinarily be considered evidence of ‘extensive’ activity are dealt with elsewhere in the Guidelines. For example, in fraud cases, the base offense level can be raised according to the amount of loss, the extent of planning, and the number of victims. [U.S.S.G] § 2F1.1. Further adjustments can be made according to the vulnerability of the victim, § 3A1.1, the defendant’s role, §§ 3B1.1, 3B1.2, and abuse of a position of trust, § 3B1.3.
105 F.3d at 802.
Wliile our dissenting colleague correctly points out that the Guidelines’ key distinction for double counting is between permissible and impermissible, see Dissent at 53-54, the cited case, United States v. Valdez-Torres, 108 F.3d 385, 389 (D.C.Cir.1997), holds simply that the prospect of double counting does not allow a court to “ignore the plain language of the Guidelines.” Id. That is not inconsistent with joining the Second Circuit in taking duplication into account in the construction of an ambiguous phrase such as “otherwise extensive.”
Of course, a court could address the concern for impermissible double counting by finding extensiveness only in characteristics (besides the number of actors) not adequately taken into account elsewhere in the Guidelines. But the Sentencing Reform Act expressly contemplates enhancement for such omissions or underassess-ments, allowing the sentencing court to depart from the otherwise applicable range if “the court finds that there exists an aggravating or mitigating circumstance of a kind, or to a degree, not adequately taken into consideration by the Sentencing Commission in formulating the guidelines that should result in a sentence different from that described.” 18 U.S.C. § 3553(b); see generally Koon v. United States, 518 U.S. 81, 91-96, 116 S.Ct. 2035, 135 L.Ed.2d 392 (1996). So it seems probable that any straying beyond the number of persons involved would at best create two headings under which these extras would be considered, and at worst cause unauthorized double counting.
The government suggests that we have' already implicitly rejected the Second Circuit view, pointing to United States v. Sobin, 56 F.3d 1423, 1428 (D.C.Cir.1995), and United States v. Dale, 991 F.2d 819, 857 (D.C.Cir.1993). In Sobin, we did rely on factors other than head count to uphold an “otherwise extensive” determination, but in that case the defendant challenged only the factual findings, not the legal standard. 56 F.3d at 1428. Dale similarly did not involve a conflict as to the standard, and in fact the trial court had found at least five persons involved. 991 F.2d at 857. Thus we are free to adopt, and do adopt, the view of the Second Circuit that § 3Bl.l(a) is “not so much about extensiveness in a colloquial sense as about the size of the organization in terms of persons involved that a defendant ‘organize[d]’ or ‘le[d].’ ” Carrozzella, 105 F.3d at 803.
We further agree with the Second Circuit that, at a minimum, “ ‘Section 3Bl.l’s ‘otherwise extensive’ prong demands a showing that an activity is the functional equivalent of an activity involving five or more participants.’ ” Id. at 803 (quoting United States v. Tai, 41 F.3d 1170, 1174 (7th Cir.1994)). To read the “otherwise extensive” prong as a lesser requirement would either allow this provision to eat up the “five or more participant” prong or would produce the anomalous result that unknowing outsiders count more than criminally culpable participants. See id. A necessary implication of this analysis is that the number of persons involved must total at least five, as it is hard to see how any lesser number could constitute the functional equivalent of five or more knowing participants. But what is necessary may not always be enough.4
[50]*50In Carrozzella the Second Circuit observed that the use of “unknowing participants”5 may be “less efficient” than the use of knowing participants or “may still only minimally further the criminal activity.” 105 F.3d at 804. Cf. Tai, 41 F.3d at 1174-75 (“If a district court intends to rely solely upon the involvement of a given number of individuals to support a determination that criminal activity is ‘otherwise extensive,’ it must point to some combination of participants and outsiders equaling a number greater than five.”). To ensure that the “extent of harm and degree of culpability in organizing or leading five unknowing participants” is not less than in case of knowing participants, Car-rozzella permits the sentencing court to “take into account the role and performance as well as the number of unknowing participants.” 105 F.3d at 804 (emphasis added).
There remains the issue of who should be counted once we include those who are unknowing or otherwise not criminally involved. Indeed, the government claims that Wilson’s enhancement is justified even under the Catrozzella framework. The Second Circuit considered the problem, noting as an example that it was necessary to distinguish the “taxi driver who brought a leader of the fraudulent scheme to work on a single occasion” from the “[salespeople who unknowingly conveyed fraudulent misrepresentations at a defendant’s request.” Id. The court identified the following factors as relevant to the head count:
(i) the number of knowing participants; (ii) the number of unknowing participants whose activities were organized by or led by the defendant with specific criminal intent [as opposed to mere service providers]; and (iii) the extent to which the services of the unknowing participants were peculiar and necessary to the criminal scheme [rather than fungible with others generally available to the public].
Id. at 803-804. We agree that these criteria are relevant. '
In trying to fit the present case under Carrozzella, the government claims that “[d]ozens of ‘unknowing participants’ were involved, carrying out [Wilson’s] bogus directions to open accounts and change addresses without realizing that [Wilson] was not who he claimed to be.” Appellee’s Brief at 47. In the government’s view, the role of these “unnamed bank employees” is analogous to that of the hypothetical salespeople mentioned in CarTozzella, who convey fraudulent misrepresentations on behalf of the defendant.
But even if we assume that the activities of the various bank personnel invoked by the government were “peculiar and necessary” to Wilson’s scheme, we do not see how they could be described as “organized or led” by him. The bank employees who changed account addresses and issued credit cards on Wilson’s instructions were, so far as appears, simply performing routine tasks that, according to pre-estab-lished bank policies, followed automatically once Wilson provided the necessary information. Such automatic behavior by functionaries of a victim institution appears totally different from the salesmanship of people retained by the defendant to market and sell a product with representations that, unbeknownst to the agent, are false. To hold otherwise would risk the absurdity that a defendant who procured 10 different [51]*51credit cards on 10 different days would escape upward adjustment if the same telephone operator happened to receive his request each time, whereas an otherwise identical defendant would get the enhancement solely because each of his calls connected him to a new employee. To take the government’s logic one step further, upward adjustment might turn on the complexity of a bank’s internal bureaucracy, with the count of unknowing participants determined by the number of desks over which a fraudulent request must pass.
The government alternatively suggests, presumably as a as a matter of common sense, that we can infer that Wilson must have used a number of additional knowing confederates. But it does not point to evidence of bank operations from which we (or the district court) could reasonably draw such an inference.
Finally, we note that in United States v. Nolan, 136 F.3d 265, 273 (2d Cir.1998), the Second Circuit applied the Carrozzella test without discussion of whether the unknowing participants were in any way organized or led by the defendant. Of course we cannot say whether this foreshadows a Second Circuit retreat from Carrozzella, but in any event we think Carrozzella got it right.
Accordingly, we see no basis for the four-point enhancement under § 3Bl.l(a). We vacate the sentence and remand the case for further proceedings consistent with this decision. As noted earlier, we remand the claim of ineffective assistance of counsel. The conviction is otherwise affirmed.
So ordered.