dissenting in part.
If anyone doubted that the Chicago Outfit during its heyday ranked as one of the most dangerous and reprehensible criminal organizations in our nation’s history, the record compiled in this case would put those uncertainties to rest. And the five defendants now before us— Frank J. Calabrese, Sr., James Marcello, Joseph Lombardo, Paul Schiro, and Anthony Doyle — sat at the very top of the enterprise. The indictment on which this quintet stood trial is breathtaking in its temporal and substantive scope: through the convenient device of the conspiracy offense, the government has been seeking to hold the defendants responsible for virtually everything that the Outfit did or sponsored for a 42-year period (1960— 2002). Although I have a few reservations that I explain below about the convictions of Lombardo, Schiro, and Doyle, in the end I join my colleagues in affirming their convictions and sentences. Regrettably, however, I must part company with their assessment of the double jeopardy argument that Calabrese and Marcello have advanced. In their view, ante at 527-28, that argument is “even weaker” in light of the evidence presented at the second trial than it was when this panel rejected this argument before the 2005 trial began. See United States v. Calabrese, 490 F.3d 575 (7th Cir.2007). I draw the opposite conclusion: the double jeopardy violation that I feared would occur from this retrial has unequivocally occurred. Calabrese and Marcello had each already been convicted and imprisoned for their part in the street crews that lie at the heart of the Outfit’s Chicago operation. See United States v. Zizzo, 120 F.3d 1338 (7th Cir.1997) (Marcello), and ante at 525. Those prosecutions covered the period from 1978 to 1992 for Calabrese and from 1979 to 1990 for Marcello. The current prosecution entirely subsumes the span of those conspiracies. I therefore dissent, on that basis only, from the decision to affirm those two convictions.
I
At first glance, the Fifth Amendment’s prohibition that no person can be “twice put in jeopardy of life or limb” for “the same offense,” U.S. Const. Amend. V, is clear enough. As we have explained, “the double jeopardy clause imposes limits on a defendant’s criminal exposure.... [T]he government cannot reprosecute a defendant for the same offense whenever it obtains broader evidence of criminal culpability.” United States v. Thornton, 972 F.2d 764, 765 (7th Cir.1992). But this simple rule becomes difficult when the “same offense” in question is a conspiracy; the problems compound when it is a RICO conspiracy. A conspiracy has “no easily discernable boundaries with regard to time, place, persons, and objectives.” Id. How, then, can we tell when one conspiracy ends and another picks up? The question becomes even more vexing when we deal with members of a complex enterprise who have allegedly conspired to violate RICO. A RICO “enterprise” is loosely defined as “a group of persons associated together for a common purpose of engaging in a course of conduct.” United States v. Turkette, 452 U.S. 576, 583, 101 S.Ct. 2524, 69 L.Ed.2d 246 (1981).
This case requires us to decide under what circumstances it is permissible to carve multiple “enterprises” out of one group. And we must do so with reference to the tightly organized, hierarchical organization commonly known as the Chica-
*536go Outfit. As the prosecution has conceded, the Chicago Outfit was organized as follows between 1960 and 2005:
[[Image here]]
Each Street Crew was headed by its own Boss, called a “Capo” (literally meaning “head,” from the Latin word “capus”— familiar to English speakers from the word “decapitate,” meaning to cut off the head). As I will describe in more detail in a moment, it is true that the earlier prosecutions of Calabrese and Marcello focused primarily on their work at the Street Crew level than on the relation between the Crews and the Boss, while the current cases look at the big picture. But that does not change the fact that both cases are inescapably about the entirety of the operation. Tempting though it may be to slice these activities more finely when we evaluate the earlier cases (pretending that the Street Crews were somehow independent of the higher echelons of the organization) and to focus on the vertical relation between the Boss and the Crews (pretending that the organization as a whole had some existence apart from its Street Crews), the facts compel the conclusion that those inside and outside the group understood throughout every relevant time that this was all one integrated, highly coordinated organization.
The majority has drawn an analogy to complex legitimate corporate enterprises (which obviously should be no worse off under either RICO or the Double Jeopardy Clause than their illicit counterparts), but this exercise does not strengthen its point. Suppose we think of the Outfit as a company and the Street Crews as its branch offices, rather like the Ford Motor Company and its River Rouge Complex. The majority concedes that a worker at the Ford River Rouge Complex is affiliated not only with that immediate Complex, but is in fact an employee of the overarching enterprise known as Ford Motor Company. Ante at 525-26. By working on the assembly line there, he contributes to Ford’s business. Id. And if Ford made two different products — say cars and bicycles (or sawed-off shotguns, as the majority postulates) — the worker on the car line would still be working for Ford, just as the worker in the bicycle plant (or the shotgun business) would be. The key point is that there is only one enterprise, which makes money through multiple lines of commerce.
The majority notes that certain actions can be taken by the line workers (the Crew members) only with the approval of central management (the Boss). In the Outfit’s case, this includes committing murder; in the Ford example, we can imagine a host of more mundane activities such as deciding to build a new line of cars, making a hiring decision, or authorizing an expenditure over $1,000. Such limitations on the authority of lower management and line workers are routine in the business world; no one subject to them
*537would think for a moment that the actions he was authorized to take on his own (such as expenditures below the threshold) were not for the enterprise’s welfare, while actions he took with approval of higher management were. The Ford employee is still a Ford employee, whether he exercises delegated discretion or whether he must follow the orders of his Ford superiors. Should the janitorial staff at the River Rouge Complex be considered to be conspiring with a different “enterprise” than a notional enterprise made up of the assembly line workers? What if the sanitation workers required approval from HR before they hired a new janitor to join their ranks? Would the action of hiring a janitor somehow become associated with the “HR-enterprise,” but all other janitorial actions remain confined to the “janitor-enterprise”? Nothing in either the Double Jeopardy Clause or RICO calls for such inconsequential distinctions. Indeed, if the majority’s view were correct, we would eviscerate any protection the Double Jeopardy Clause provides against repeat prosecutions for conspiracy; single organizations could be carved into any number of different “enterprises” to avoid the Clause’s protection. (I note in passing that the Supreme Court has treated corporations as “persons” for purposes of the Double Jeopardy Clause. See, e.g., United States v. Martin Linen Supply Co., 430 U.S. 564, 97 S.Ct. 1349, 51 L.Ed.2d 642 (1977).)
To make the analogy clearer, let’s pretend that a hypothetical car manufacturer, Voiture, is using some of its employees to run a video poker side-business at a local bar, and that the employees are well aware that these activities violate the law. Let’s further assume that a Voiture employee works full-time at its assembly line in Indiana, spending most of her days at that facility making cars but occasionally conferring about the poker business with her superiors at headquarters over the phone or in person. Law enforcement agents get wind of illegal conduct taking place and bring an indictment against the Indiana employee. The indictment charges that the Indiana facility is a RICO enterprise, and that the employee has conspired with members of that enterprise to further the activities of the video poker business at the bar in question, using company facilities and time. After a trial, a jury finds her guilty and she serves time in prison.
Years pass, and another Voiture employee decides that he has had enough with the corporation and its illegal activities. He decides to turn on his coworkers and tell law enforcement everything he knows. (Or perhaps, closer to this case, confronted with his own misdeeds he comes clean in exchange for the government’s leniency.) Through this informant, officials have proof for the first time that the employee who was prosecuted earlier actually was handling video poker for Voiture in all of central Indiana, not just in the bar that was involved in the first case. They decide to charge her again, this time with an indictment covering the full scope of her crimes. Again, rather than charge her for the underlying substantive conduct, they charge her with conspiracy. This time, prosecutors are careful to say that the enterprise is Voiture as a whole, not just the Indiana regional center. Moreover, they emphasize that Voiture’s central management had to approve each location for the illegal machines. This, they say, avoids any double jeopardy problem, because (the argument goes), the enterprise whose illegal activities she was furthering the second time was Voiture, not its Indiana plant.
Such a distinction would be absurd. Higher management was already fully implicated in the earlier scheme. Nothing separates the “enterprise” of the plant *538from the “enterprise” of the company as a whole. Companies work through people; large companies usually find it convenient to work through divisions based on geography, line of business, or both. The Indiana employee, by working for the Indiana assembly plant and consulting as need be with higher management, was by definition working for the company as a whole. The fact that Voiture organizes itself in a vertical structure with regional manufacturing centers does not mean that each center is a separate enterprise from Voiture itself, even if the centers cannot take certain actions without the approval of Voiture’s management. This reality cannot be evaded by naming the regional center of Voiture in the charging documents the first time around. Nor, in this case, can it be evaded by naming the Street Crews first and later appealing to an Outfitwide conspiracy.
Returning to our case, no one disputes the fact that the Calabrese and Marcello Street Crews operated within and exclusively for the Outfit. This can only mean that their prosecutions were for the work that they did for the Outfit, each one through his own Street Crew. The facts developed at trial simply do not support the proposition that the Crews were standalone operations, acting as independent contractors for the Outfit. Nor is this a case in which either Calabrese or Marcello is being asked to be criminally responsible for the activities of other Street Crews, qua Street Crews. The only difference between the present case and each man’s earlier prosecution — a difference to which the government alludes repeatedly — is the wider scope of the recent prosecution, and especially the fact that it encompasses murders authorized at the highest levels of the Outfit. Disturbing though this conduct is, however, these murders do not support the proposition that the enterprise known as the Outfit is different from the enterprises involved in the first cases. We must recognize, as have our sister circuits, that a crime family in “a lower level of authority within the hierarchy of organized crime” is still a component of the same crime family. United States v. Langella, 804 F.2d 185, 189 (2d Cir.1986); see also United States v. Ciancaglini, 858 F.2d 923 (3d Cir.1988) (concluding that two Philadelphia-based crime families were part of the same enterprise). If the Street Crews were “self-sufficient enterprises that function!] without oversight” from the Outfit, we would have a different case. Langella^ 804 F.2d at 189. But as the majority concedes, they are not. The Street Crews were the mob’s hands, the Outfit its head. There is no way to divide the two.
II
My dissent does not proceed from the assumption that one person is incapable of entering into two different RICO conspiracies with the same enterprise. I agree with the majority that the contrary is true. As the Second Circuit noted in United States v. Basciano, “enterprise and pattern are distinct elements of racketeering.” 599 F.3d 184, 204 (2d Cir.2010). I therefore have no quarrel with the proposition that a person who has once been prosecuted for a low-level conspiracy (perhaps to sell marijuana from a corrupt branch office of a company), is not immune from prosecution in a different, much larger conspiracy (such as a nationwide conspiracy orchestrated at the highest levels to commit financial fraud). In that example, even though the wrongdoer would have made a second agreement with the same enterprise, it would have been an agreement to commit a different pattern of racketeering activity.
As I read the majority’s opinion, it accepts that if the Carlisi and 26th Street *539Crews were doing the actual work of the Outfit during the times covered by their earlier indictments, then this would be a different case. But, they conclude, neither Crew was doing so. My problem with that conclusion is not with the theory but with the application. As I said before, it is certainly possible that a case could arise in which actions taken by the Outfit amounted to a different pattern of racketeering than the activities that take place at the Crew level, even though the two are part of the same enterprise. But the facts of this case show instead a single coordinated operation. We can see this by considering the various types of evidence that shed light on the question whether two conspiracies conducted by the same enterprise are distinct. This includes “(1) the time of the various activities charged as separate patterns of racketeering; (2) the identity of the persons involved in the activities under each charge; (3) the statutory offenses charged as racketeering activities in each charge; (4) the nature and scope of the activity the government seeks to punish under each charge; and (5) the places where the corrupt activity took place.” United States v. Marren, 890 F.2d 924 (7th Cir.1989); see also United States v. Sertich, 95 F.3d 520 (7th Cir.1996). When the answers to each of these questions point in the same direction, the court must find that there is just one pattern of racketeering and the conspiracies had essentially the same object. In such a case, it would violate double jeopardy to bring a second prosecution.
It may help in this case to compare the first and second prosecutions using a table, beginning with Calabrese’s case. It is undisputed that the time and location of his earlier indictment are completely subsumed within the present one. I thus focus on the parts of the indictment summarizing the offenses charged:
Calabrese 1995 Indictment 2005 Indictment
Enterprise “The Calabrese Street Crew was part of a “The Chicago Outfit was known to its larger criminal organization known to the members and associates as ‘the Outfit’ and public as ‘the Mob,’ and to its members and was also known to the public as ‘organized associates as ‘The Outfit.’ ” crime,’ the ‘Chicago Syndicate’ and the ‘Chicago Mob.’ ”
Purpose “The Calabrese Street Crew existed: (1) to “The Chicago Outfit existed to generate generate income for its members through income for its members and associates illegal activities, and (2) to cover up and to through illegal activities.” conceal evidence of the crew’s involvement in illegal activities after commission of those illegal acts.”
Activities “The illegal activities of the crew included, but were not limited to: (1) making loans to individuals at usurious rates of interest [juice loans] ... (2) ‘collecting’ through ‘extortionate means’ juice loans constituting ‘extensions of credit,’ ... (3) collecting debts incurred in the crew’s juice loan business, ... (4) using threats, violence and intimidation to collect juice loan debts and to discipline crew members; (5) devising a scheme to defraud and to obtain money and property by means of false and fraudulent representations through the use of the mails; and (6) tampering with witnesses to, and victims of, the crew’s illegal activities.” “The illegal activities of the Chicago Outfit included, but were not limited to: (1) collecting ‘street tax,’ that is, extortion payments required as the cost of operating various businesses; (2) the operation of illegal gambling businesses, which included sports bookmaking and the use of video gambling machines; (3) making loans to individuals at usurious rates of interest [juice loans]; (4) ‘collecting* through ‘extortionate means’ juice loans constituting ‘extensions of credit’ ... (5) collecting debts incurred in the Chicago Outfit’s illegal gambling business ... (6) collecting debts incurred in the Chicago Outfit’s juice loan business ... (7) using threats, violence, and intimidation to collect street tax and juice *540loan debts; (8) using threats, violence, and intimidation to discipline Chicago Outfit members and associates; (9) using murder of Chicago Outfit members, associates and others to advance the interests of the Chicago Outfit’s illegal activities; (10) obstructing justice and criminal investigations by ... murdering witnesses ... and (11) traveling in interstate commerce to further the goals of the criminal enterprise.”
These quotations from the two indictments demonstrate that the only difference between the earlier and the later one is that the second contains a wider array of alleged criminal activity. But federal courts use a “same offense” test for double jeopardy purposes, not a “same evidence” or even a “same allegation” test. United States v. Dixon, 509 U.S. 688, 696, 113 S.Ct. 2849, 125 L.Ed.2d 556 (1993). Thus, if the pattern of activity is the same, even if there are some difference in detail, this points to a finding of “same offense.”
Here, the second indictment adds to the first’s list of the Outfit’s illegal activities and in some respects is more specific. It offers more detail about the street tax and illegal gambling operations, and it squarely accuses the defendants of committing murder in furtherance of their illegal conspiracy. Obviously, murder is as serious a charge as can be made, but the addition of murder to the list does not change the nature of the offense with which these defendants were charged: RICO conspiracy. Although the government and majority focus on murder as the key distinguishing feature, they overlook the fact that the earlier indictment accused Calabrese of being responsible for highly violent activity against both Outfit members and witnesses. When the federal government later uncovers additional evidence of discrete acts of such violence, it is free to prosecute Calabrese for those acts (assuming that a federal statute covers them) or assist state authorities in a state prosecution, but it cannot reprosecute him for the agreement he made with the Outfit to engage in that pattern of conduct just because it finds evidence of ever more heinous actions in support of that agreement. It is worth noting that if the earlier charge had been a substantive one accusing Calabrese of extortion, and the new indictment charged him with the substantive offense of murder, the situation would be entirely different: those are two different offenses. Indeed, the government might this time around have been able to prosecute one or both of these defendants for conspiracy to commit murder for the purpose of gaining entrance to or maintaining a position in an enterprise engaged in racketeering, in violation of 18 U.S.C. § 1959(a). See Basciano, 599 F.3d at 198-99 (holding that a conspiracy to violate Section 1959 is not the same offense as a conspiracy to violate Section 1962). But that is not the choice that it made.
The overlap in Marcello’s two indictments is even greater. Calabrese’s second indictment differed slightly from the first because it contained more detailed references to illegal gambling, street tax, and the additional allegations of specific murders. Marcello’s first indictment is even closer to the second because the first referred to illegal gambling and attempted murder. And because Marcello went to trial in both the earlier and present cases, the evidence presented at his trials brings the double jeopardy violation into even sharper view. Crucially, given the majority’s current emphasis on the murder evidence, the government also presented evi*541dence about the commission of six murders at Marcello’s first trial. Finally, in both of Marcello’s trials the government elicited testimony that implicated the same nineteen people (in addition to the five standing trial) in the Outfit’s conspiracy. The current trial had an unmistakable air of déja vu.
The majority may well be correct that its hypothetical Ford worker who agrees to manufacture guns at time A could also be convicted of a separate conspiracy if, at time B, he agrees to work at corporate headquarters to conceal the illegal income from guns. Ante at 526. To determine whether those two prosecutions would be barred by the Double Jeopardy Clause we would look to the same five-factor test outlined above; if the activities were indeed sequential and did not overlap and the activities were distinct as they seem to be (building guns versus concealing income), there may be no problem with prosecuting the income-concealment conspiracy after gun-manufacturing conspiracy. Unfortunately, that example does not describe this case. Here, the government’s charges against Marcello and Calabrese covered the same period of time and the same pattern of racketeering activity. The Outfit’s commission of violence and murder was a greater focus of the government’s case the second time around, but it was also a component of the first two prosecutions.
Perhaps the government played its cards too soon by moving ahead with the earlier prosecutions (how could it have known that in 1999 the FBI would rediscover gloves that Nick Calabrese carelessly discarded after the 1986 Fecarotta murder, that the gloves would still have Nick’s DNA on them, and that this would lead him to flip), but that is the price that occasionally is exacted by the Double Jeopardy Clause. Conspiracy can reach back almost indefinitely. If the conspiracy itself is a durable one that lasts over many years or even decades, as this one did, the indictment could (as this one did) reach back even to the year in which the distinguished U.S. Attorney for the Northern District of Illinois was born. When the government chooses to use this broad and powerful tool once, however, “its choice has consequences.” Baseiano, 599 F.3d at 203. One of those consequences is refraining from prosecuting the defendant again, for the same conspiracy, when it obtains broader evidence of criminal culpability. As I explained in my separate opinion before these trials went forward, I see no difference in the essential agreement that was at issue in the earlier cases and in this case. I would reverse Calabrese and Mar-cello’s convictions on the ground that the present trial has violated their rights under the Double Jeopardy Clause. To this extent, I therefore respectfully dissent.
Ill
Although I agree with the outcome the majority reaches on the remaining issues, I find two of those questions to be closer than they do, and so I add a few words about them.
A. Voir Dire
All of the defendants except Doyle argue that the district court should have asked the jurors whether they had been exposed to various news articles that were published during the trial. I agree with my colleagues that the district court’s decision not to do so does not amount to reversible error. Even if a district court’s failure to voir dire is error, we reverse only if “there is any substantial likelihood that the defendants were denied a fair trial.” United States v. Balistrieri, 779 F.2d 1191, 1214 (7th Cir.1985). Here, even if the jurors had read all of the items the defendants *542have complained about, it would not have made any difference to them. Things might be different if the news articles had contained references to inadmissible evidence or information going beyond the horrific account to which the jurors were exposed during the trial, but I am satisfied that those problems did not arise.
My concern is over the district court’s wholesale refusal to explore the jurors’ exposure to outside publicity. My colleagues find no problem with that and so do not need to reach the issue of harmless error; I am not so sure. When a defendant’s notoriety “guarantee[s] extensive press coverage,” ante at 530, it is imperative that the court be ready to make use of the limited two-step voir dire process we established in Margoles v. United States, 407 F.2d 727, 735 (7th Cir.1969), to ensure that the trial is fair. Voir dire helps to guarantee that a trial’s outcome is determined by events inside the courtroom, not what is going on outside in the court of public opinion. Since Margóles, we have repeatedly told district courts that when “prejudicial publicity is brought to the court’s attention during a trial ... the court must ascertain if any jurors who had been exposed to such publicity had read or heard [it].” United States v. Trapnell, 638 F.2d 1016, 1022 (7th Cir.1980) (emphasis added). This is not meant to be a burdensome procedure; only when a juror admits that she has read or heard the item in question must the court go on to examine that juror about the publicity’s effect. Id. Far from insulting the jurors, asking a simple question about whether they have read or heard an item reiterates the importance of the court’s instruction to avoid the news, and thus communicates to the jury the court’s respect for the fair trial rights of the accused. For a court to refuse to conduct voir dire even once in the course of a sensitive and lengthy trial with extensive media coverage, especially after defendants brought to light some articles that were borderline prejudicial (such as the op-ed telling jurors they were “stupid” if they did not convict), was a move that could have undermined the whole trial. A court should not risk jeopardizing the outcome of the trial by failing even to check that jurors were following the instructions. The fact that the gamble worked here, and that the record does not support a finding of prejudicial error, is not enough to commend this practice.
B. Marcello’s Voice Identification Expert
Finally, I do not believe the district court’s decision to exclude expert testimony on the reliability of voice identification evidence was correct, although I agree with my colleagues that it does not require reversal.
Marcello was accused of murdering Michael Spilotro. Spilotro’s daughter, Michelle, testified that on the day of her father’s murder, a man called their home and asked to speak to him. She testified that the same person had regularly called her father. Three years after Spilotro’s death, Michelle listened to a “voice lineup” put together by the FBI. The first five voices on the tape were those of officers reading a sample piece of text; the last was Marcello’s. Michelle picked Marcel-lo’s voice as the one she remembered hearing on the day of her father’s death. At trial, she told the jury that she was “100 percent sure” it was Marcello’s voice she had heard on the phone.
Marcello sought to have an expert, Dr. Daniel Yarmey, testify about the reliability of voice identification. Dr. Yarmey is a professor of psychology who has conducted extensive research in the areas of memory; he has investigated voice identification in particular. His testimony would have *543done much more than tell jurors “voice identifications frequently are mistaken.” Ante at 529. He was prepared to educate the jury about error rates associated with voice identification — in some studies, misidentification rates were as high as 45%— and the factors that affect the reliability of voice lineups. Dr. Yarmey had also conducted his own evaluation of the lineup that Michelle Spilotro had heard. He recruited 157 undergraduates at his university to listen to the lineup, evaluate it using a number of factors, and try to identify the suspect’s voice. The listeners were able to do so at a rate that exceeded pure chance.
The district court refused to admit this expert testimony, not because of any deficiencies in Dr. Yarmey’s qualifications, but because the district court believed that this information was something the “jury knows anyway.” The court also assessed the voice lineup on its own and concluded that there was “nothing about the difference [between Marcello’s voice and the others] that would suggest to a hearer, to a listener, that one or the other was actually the suspect.”
Even though our review of a district court’s decision not to admit expert testimony is deferential, see United States v. Carter, 410 F.3d 942, 950 (7th Cir.2005), in my view the district court’s refusal to admit Dr. Yarmey’s testimony was a mistake. In recent years, courts have become more aware of the reality that human memory is not necessarily reliable. A study of 200 wrongful convictions revealed that 79% rested in part on mistaken eyewitness identifications. Brandon L. Garrett, Judging Innocence, 108 Colum. L.Rev. 55, 60 (2008). This does not mean that courts must impose a blanket ban on such testimony, but it is critical to be cautious. We cannot ignore the power that a witness’s claim to be “100% sure” may have on a jury, nor can we ignore that such witnesses are sometimes, unfortunately, mistaken. The Supreme Court recently emphasized that one tool that courts can use to ensure juries do not give such testimony more weight than it is worth is to allow “expert testimony on the hazards of eyewitness identification.” Perry v. New Hampshire, — U.S.-, 132 S.Ct. 716, 729, 181 L.Ed.2d 694 (2012). As Dr. Yarmey’s research shows, a witness’s voice memory is not exempt from the sort of problems that we more commonly associate with a witness’s vision; just as with eyewitness identification, expert testimony on the reliability of voice identification reveals vulnerabilities that lie outside the range of common knowledge.
The district court’s decision not to admit Dr. Yarmey’s testimony evinces a misunderstanding of the purpose of expert testimony on the reliability of a witness’s memory. As we explained in United States v. Bartlett, expert testimony should not be kept out simply because a court believes “jurors know from their daily lives that memory is fallible.” 567 F.3d 901, 906 (7th Cir.2009). That may be true, but “[t]he question that social science can address is how fallible,” id., and thus how deeply the jury might wish to discount any given identification. “That jurors have beliefs about this does not make expert evidence irrelevant; to the contrary, it may make such evidence vital, for if jurors’ beliefs are mistaken then they may reach incorrect conclusions. Expert evidence can help jurors evaluate whether their beliefs about the reliability of eyewitness testimony are correct.” Id. As is clear from the district court’s remarks in this case, the court itself held beliefs about the reliability and suggestiveness of the voice lineup that are belied by the expert’s conclusions. As far as we know, the jurors shared these misconceptions. This case thus highlights why it is critical for jurors to hear expert testimony in order to be able correctly to evaluate a witness’s mem*544ory. Just because courts have routinely admitted laywitness identification in the past is no reason to continue to do so without skepticism, in light of modern research showing the fallibility of such identifications. When a court does admit such identification testimony, expert testimony will often be necessary to enable jurors to properly evaluate its reliability.
I do not believe, however, that this error warrants reversal of Marcello’s conviction. Even if Michelle Spilotro had not testified, there was ample additional evidence — notably Nick Calabrese’s testimony — that implicated Marcello in Spilotro’s murder. The error was therefore harmless.
In conclusion, I would affirm (with the minor adjustment for Doyle’s restitution obligation discussed in the majority’s opinion) the convictions and sentences of Joseph Lombardo, Paul Schiro, and Anthony Doyle. I would reverse the convictions of Frank J. Calabrese, Sr., and James Mar-cello, on the ground that this prosecution violated each man’s rights under the Double Jeopardy Clause. To that extent, I respectfully dissent.