James I. Glasser
In many investigations and prosecutions involving concerted criminal activity, a paramount objective is to identify a “cooperator” – a witness who can identify the criminal actors and narrate the criminal activity for the benefit of the trier of fact. Oftentimes, the more involved in the criminal activity, the more attractive the individual is as a cooperator.
It is hard to imagine what a person would be willing to do or say when facing a long prison term and knowing that the only way to mitigate the sentence by half or more is by currying favor with law enforcement authorities and by inculpating others. What tales would a person be willing to tell to win freedom?
To be sure, cooperating witnesses play an important role in our criminal justice system. As the standard jury instruction in the Second Circuit provides:
. . . only people who themselves take part in criminal activity have the necessary knowledge to show criminal behavior by others. For those reasons, the law allows the use of accomplice testimony. Indeed, it is the law in federal courts that the testimony of accomplices may be enough in itself for conviction.
Approximately 75 years ago, in United States v. Dennis, 183 F.2d 201, 224 (2d Cir. 1950) aff’d, 341 U.S. 494 (1951), Judge Learned Hand recognized this reality when he observed:
Courts have countenanced the use of informers from time immemorial; in cases of conspiracy, or in other cases when the crime consists of preparing for another crime, it is usually necessary to rely upon them or upon accomplices because the criminals will almost certainly proceed covertly.
Without cooperating witnesses, much criminal activity would never be ferreted out. But what safeguards are there to ensure that the testimony of a cooperator is reliable and truthful? There is no gating mechanism to screen cooperator testimony, similar to a Daubert hearing. The reality is that determinations of reliability rest with the prosecutor alone. Juxtaposed against this reality is the fact that the promise of leniency is a powerful motivator ‒ a motivator perhaps to shade the truth, exaggerate, or even provide blatantly false testimony.
With the foregoing in mind, it is important to recall that under our system of justice, a party to a controversy may not pay or give anything of value to a witness in exchange for that witness’ testimony. How then to reconcile the exchange of testimony for a promise of lenience? An important exception to this general rule permits a prosecutor to exchange a promise of lenient treatment to criminals in exchange for their testimony that assists in the investigation and/or prosecution of others. This is a carrot and stick arrangement – the cooperator is always under the threat of prosecution for false statements or perjury for giving false testimony, and the reward is conditioned on truthful and complete testimony. But the reality is that the stick is rarely wielded. There are precious few reported instances of cooperators being prosecuted for making false statements or for perjury. On the other hand, there is no shortage of defendant exonerations owing to a finding that perjured testimony contributed to the conviction.
Judges have long been sensitive to the risks inherent in using and rewarding cooperating witnesses with leniency in return for their testimony. See generally, Ellen Yaroshefsky, Cooperation with Federal Prosecutors: Experiences of Truth Telling and Embellishment, 68 Fordham Law Review 917 (1999).
In 1837, Lord Abinger, the nineteenth-century English politician and judge, wrote: “[T]he danger is that when a man is fixed, and knows that his own guilt is detected, he purchases impunity by falsely accusing others.”
This grave concern has been consistently expressed by judges throughout the centuries. More recently, in 1996, Judge Easterbrook of the U.S. Court of Appeals for the Seventh Circuit noted the special danger of using a cooperator who perceives lenience to be dependent on the cooperator’s performance on the witness stand. Noting that “[i]nformants are tempted to manufacture or exaggerate evidence of crime,” Judge Easterbrook observed in United States v. Cook, 102 F.3d 249, 252 (7th Cir. 1996), that:
[a] turncoat (that is, the defendant’s former associate-in-crime who turns state’s evidence) is not monitored [by wire] and therefore has greater leeway for invention at trial. The turncoat’s reward structure . . . differs [from a wired informant]. Like defendants, turncoats see performance on the stand as the key to less time in prison. Compensation contingent on testimony favoring the prosecutor may create pressure to lie. . . .
In 2001, in Commonwealth of N. Mariana Islands v. Bowie, 243 F.3d 1109, 1124 (9th Cir. 2001), the U.S. Court of Appeals for the Ninth Circuit summarized its attitude of skepticism toward cooperators’ testimony as follows:
[B]ecause of the perverse and mercurial nature of the devils with whom the criminal justice system has chosen to deal, each contract for testimony is fraught with the real peril that the proffered testimony will not be truthful, but simply factually contrived to “get” a target of sufficient interest to induce concessions from the government. Defendants or suspects with nothing to sell sometimes embark on a methodical journey to manufacture evidence and to create something of value, setting up and betraying friends, relatives, and cellmates alike.
In addition to judicial observations about dangers inherent in the use of cooperating witnesses in the prosecution of others, there is no shortage of empirical data confirming those observations. The National Register of Exonerations Annual Report for 2024 reports that there were 147 recorded exonerations that year. Of those, 106 exonerations – 72 % – were attributable to demonstrated perjury or other false accusations. See www.exonerationregistry.org.
Similarly, a 2005 study by the Center for Wrongful Convictions at Northwestern School of Law examined 111 exonerations in capital cases and found that in 51 of 111 cases, defendants were wrongfully sentenced to death based at least in part on the testimony of “witnesses with incentives to lie,” making “snitches,” according to the study, “the leading cause of wrongful convictions in U.S. capital cases.” Center on Wrongful Convictions, The Snitch System: How Snitch Testimony Sent Randy Steidl and other Innocent Americans to Death Row, 3, available at http://www.innocenceproject.org/docs/SnitchSystemBooklet.pdf.
With that backdrop, the use of cooperating witnesses in criminal prosecutions is well-entrenched and a long-standing practice. This article traces the historical use of conspirator testimony, the long-standing concerns with such testimony, and the recognition that the use of such witnesses may be a necessary evil inherent in any criminal justice system.
Witness Rules in the Old Testament
Scholars trace the history of cooperating witnesses to Medieval England. In truth, the need for checks on a cooperating witness dates back to the Old Testament. The Book of Numbers, Chapter 35, Verse 30, provides: “Whoso killeth any person, the murderer shall be slain at the mouth of witnesses; but one witness shall not testify against any person that he die.” Thus, according to the Old Testament, at least two witnesses to a murder are required to impose capital punishment. Deuteronomy 17:6 also specifies that at least two witnesses to a murder are required to impose a death sentence: “At the mouth of two witnesses, or three witnesses, shall he that is to die be put to death; at the mouth of one witness he shall not be put to death.”
Deuteronomy 19:15 establishes a sweeping requirement that guilt be established by two or more witnesses in all criminal and civil matters:
One witness shall not rise up against a man for any iniquity, or for any sin, in any sin that he sinneth: at the mouth of two witnesses, or at the mouth of three witnesses, shall the matter be established.
Today we see echoes of these witness corroboration rules in jurisdictions that require corroboration of cooperator testimony. While in the federal system corroboration is not required and the testimony of a co-conspirator alone is sufficient to support a conviction, many states do require corroboration. For example, in New York the rule is that a “defendant may not be convicted . . . upon the testimony of an accomplice alone unsupported by corroborative evidence tending to connect the defendant with the commission of such offense.” N.Y. Criminal Procedure Law § 60.2. Although the corroborating evidence need not be so strong that, standing alone, it suffices to convict, New York courts are clear that “if the corroboration requirement is not met, a conviction cannot stand.” People v. Reome, 933 N.E.2d 186, 189 (N.Y. 2010). Like the ancient rule requiring multiple witnesses, this modern rule provides a check on cooperator witness testimony.
The Law of Approvement
As early as 1130, the Laws of England recognized the practice of approvement. An approver was a charged person who acknowledged his guilt and simultaneously identified his accomplices in exchange for a pardon. Neal Kumar Katyal, Conspiracy Theory, 112 Yale Law Journal 1307 (2003); 2 M. Hale, The History of the Pleas of the Crown (hereafter Hale) 225-235 (1736). See also John H. Langbein, Shaping the Eighteenth-Century Criminal Trial: A View from the Ryder Sources, 50 U. Chi. L. Rev. 63, 91 (1983) (hereafter Langbein).
If the court permitted the offender to “approve,” the approver would make a detailed out-of-court confession and accusation against accomplices. Then, at the trial of his accomplices, the approver repeated his accusations. If the approver’s accomplices were convicted, the approver was pardoned.
The practice of approvement was not without its perils to the “approver.” First, a defendant could become an approver by seeking a pardon only after acknowledging his or her own guilt. Only then could a court exercise its unlimited discretion to accept or reject the would-be approver’s request to testify against the would-be approver’s accomplices. Sometimes defendants confessed their crimes in open court only to be denied status as an approver. Would-be approvers who found themselves in this position were typically executed.
Another hurdle was the fact that an approver’s pardon was contingent upon conviction of those against whom the approver was approving. If the approver’s accomplices were convicted based on the approver’s testimony, the approver was entitled to be freed. But if the approver’s accomplices were acquitted, the approver would be hanged without trial based on the approver’s previously entered plea of guilty.
Approvement also had byzantine procedural rules. Approvers who violated these rules were often executed. For example, if the approver hesitated more than a few days in making a full, out-of-court accusation against the approver’s accomplices, the approver “would be hanged.” Or if the court discovered that the approver’s accomplices had absconded by the time the approver accused them, the approver “would be hanged.” And, assuming the approver survived until trial, if the approver erred in any matter when repeating accusations at trial, including seemingly minor details, the approver would be hanged. The approver would also be hanged if the approver disavowed his accusation when brought into open court.
On the other hand, if the approver was successful in surviving this gauntlet, a complete pardon was the prize. The brass ring of a pardon of course presented a motive to falsely accuse others. Critics of approvement, including the common-law treatise writer Matthew Hale, observed that “more mischief hath come to good men by . . . approvements . . . than benefit to the public by the discovery and convicting of real offenders”
The Crown Witness System
By the mid-1700s, the statutory and common law crown witness system replaced approvement as the vehicle to encourage accomplice testimony. Langbein, 91. Like the approver, the crown witness avoided imprisonment by testifying against the crown witness’ accomplices. Lord Mansfield, the Lord Chief Justice of the King’s Bench from 1756-1758, described approvement as the “analogy to which [England’s crown witness system was] adopted.” Rex v. Rudd, 168 Eng. Rep. 160, 162 (K.B. 1775) (Mansfield, C.J.).
Unlike the approver, whose freedom was contingent on the conviction of the approver’s accomplices, the crown witness had only a “best efforts” obligation to provide testimony against accomplices. Henry Fielding, London’s Chief Magistrate in the middle of the 1700s, explained that once the crown witness testified against his coconspirators, “whatever be the Fate of his Evidence, [he] always [went] free.” Langbein, 92 (1983). Only where the crown witness acted in bad faith, or failed to testify fully and fairly, would he be prosecuted. Whiskey Cases, 99 U.S. 594, 600 (1878), citing Rex v. Rudd, 1 Leach 115, 119, 168 Eng. Rep. 160, 162 (K.B. 1775) (Mansfield, C.J.).
The crown witness system also differed from approvement in that the crown witness confessed his crimes to a magistrate at the pretrial stage, unlike the approver, who sought a pardon from the court for his testimony, only after admitting his guilt. The crown witness’ reward for his cooperation was the magistrate’s agreement not to prosecute him. See Langbein, 92 (1983). The crown witness therefore remained unindicted/uncharged throughout his accomplice’s case; he was not “pardoned” for his crime.
Many offenders took advantage of the crown witness system. According to one analysis of trials in the 1750s in London’s criminal court, the crown witness system was “practically the only resort of the London authorities” in dealing with gang crimes. Langbein, 91.
Cooperating Witnesses in the United States – the Nineteenth Century
In the United States, federal and state jurisdictions embraced the major elements of England’s crown witness practice. By the time the Whiskey Cases, 99 U.S. 594, 603 (1878), were decided in 1878, a number of American jurisdictions permitted the prosecutor to allow an accomplice to testify and thereby gain a pardon. In the Whiskey Cases, the Court seemed to favor this arrangement, noting that, compared to the trial judge, the prosecutor could better assess the relative need for the accomplice’s testimony.
The Whiskey Cases involved an agreement between a prosecutor and defendants who offered to testify against corrupt government officials. Pursuant to the agreement, the defendants agreed to plead guilty to one count of a criminal indictment, to testify fully about the conspiracy, and to withdraw their defensive pleas in a civil tax case brought by the prosecutor. In exchange, the prosecutor agreed to forego prosecution on other counts and to forego other tax claims against the defendants.
The defendants alleged that they had performed their part of the bargain, and that the prosecutor had breached the agreement between the parties by pursuing the tax claims. The Supreme Court, id. at 598, framed the issue thus:
[W]hether the district attorney, as a public prosecutor, may properly enter into an agreement with . . . an accomplice, that if he will testify fully and fairly in such a prosecution against his associate in guilt he shall not be prosecuted for the same offence; and if so, whether such an agreement, if the witness performs on his part, will avail the witness as a defence to the criminal charge in case of a subsequent prosecution.
The Supreme Court held that “the district attorney had no authority to make the agreement” and that the agreement was not a valid defense against the tax claims. Id. at 606. Drawing on its understanding of English practice, the Court explained that if a defendant chooses to testify against an accomplice, the defendant had only an equitable right to a pardon from the executive – not a legal right to enforcement of any supposed agreement with the prosecutor.
Courts in the nineteenth century generally reacted to agreements to accord leniency to cooperating defendants with disfavor. For example, the Wisconsin Supreme Court, in Wight v. Rindskopf, 43 Wis. 344, 354-55 (Wis. 1877), characterized a plea agreement in which a defendant secured leniency by pleading guilty and offering to testify against other offenders, as “hardly, if at all, distinguishable in principle from a direct sale of justice.” The court said that “[s]uch a bargain . . . could not be kept . . . in any court not willing largely to abdicate its proper functions in favor of its officers.”
But even as nineteenth-century appellate judges expressed disfavor with plea bargaining generally and bargains made with cooperating witnesses specifically, plea bargaining grew substantially, particularly in the early twentieth century. For example, in New Haven, Connecticut in 1888, 75% of all felony cases were resolved by pleas; this percentage increased to over 90% of all felony dispositions by 1921. In the federal system, in 1908, about half of all felony matters were resolved by pleas. Only eight years later, in 1916, that number increased to 72%. By 1925, the number increased to almost 90%; now, nearly 95% of federal criminal matters are resolved by pleas.
Cooperating Witnesses in the United States – the Twentieth Century
By the 1960s, the American justice system had come to embrace both plea bargaining and the prosecutorial exchange of leniency for cooperator testimony. Though the practice was embraced, make no mistake, the inherent dangers of the practice were well-appreciated. In Hoffa v. United States, 385 U.S. 293 (1966), for example, the Supreme Court considered a number of challenges to the use of a confidential informant who was present when the defendant allegedly made private, incriminating statements about an attempt to bribe a jury. Apparently in exchange for the informant’s cooperation in the bribery investigation, the government dropped or no longer pursued criminal charges that had been pending against the informant.
The Supreme Court reasoned, at 385 U.S. at 311-12, that the reliance on the informant’s testimony in the bribery trial was not inappropriate because existing safeguards for testing the accuracy and veracity of a witness, coupled with jury instructions, were a sufficient guard against a wrongful conviction based on cooperator testimony and therefore satisfied due process:
[It] is quite correct . . . that [this particular cooperator] perhaps even more than most informers, may have had motives to lie. But it does not follow that his testimony was untrue, nor does it follow that his testimony was constitutionally inadmissible. The established safeguards of the Anglo-American legal system leave the veracity of a witness to be tested by cross-examination, and the credibility of his testimony to be determined by a properly instructed jury.
Justice Warren, id. at 320, dissented from the Court’s decision:
This type of informer and the uses to which he was put in this case evidence a serious potential for undermining the integrity of the truth-finding process in the federal courts. Given the incentives and background of [this cooperator], no conviction should be allowed to stand when based heavily on his testimony. And that is exactly the quicksand upon which these convictions rest, because without [the cooperator], who was the principal government witness, there would probably have been no convictions here. . . . [T]he affront to the quality and fairness of federal law enforcement which this case presents is sufficient to require an exercise of our supervisory powers.
Justice Warren’s view was obviously the minority view and as recent statistics evidence, cooperation in exchange for leniency is alive and well and now firmly rooted in our system of justice.
Acknowledging that the cooperating witness is now firmly rooted in the American justice system does not mean that courts are not troubled from time-to-time by the dangers inherent in such witnesses. Indeed, in 1998 the U.S. Court of Appeals for the Tenth Circuit decided United States v. Singleton, 144 F. 3d 1343 (10th Cir. 1998) (Singleton I).
The facts of Singleton were hardly unusual. Sonya Singleton was convicted of money laundering and conspiracy to possess with intent to distribute cocaine. Her co-conspirator, Napoleon Douglas, entered into a plea agreement and agreed to testify in return for the government’s promise not to prosecute him for related offenses, to advise the sentencing court of his cooperation, and to advise a state parole board of the “nature and extent” of his cooperation.
Before trial, Singleton moved to suppress Douglas’ testimony on the ground the government had violated the federal anti-gratuity statute, by promising Douglas leniency in exchange for his testimony. The district court denied the motion and Douglas testified against Singleton.
Singleton appealed. The Tenth Circuit overturned Singleton’s conviction, finding that the government had in fact violated the anti-gratuity statute. That statute makes it unlawful to “directly or indirectly . . . give[], offer[], or promise[] anything of value . . . for or because of . . . testimony under oath. . . .” The relevant language is now codified at 18 U.S.C. § 201(b)(1).
Applying a plain-language reading to the statute, the Tenth Circuit determined that the statute applied with equal force to government attorneys and that the promises contained in Douglas’ agreement fell within the statutory definition of “anything of value.” The court held that such evidence should be excluded, because violations of the statute by prosecutors were “widespread and repeated,” and there was a significant likelihood of future violations to be deterred. The court, Singleton at 1359-60, explained:
No practice is more ingrained in our criminal justice system than the practice of the government calling a witness who is an accessory to the crime for which the defendant is charged and having that witness testify under a plea bargain that promises him a reduced sentence. This ingrained practice of buying testimony indicates that suppression is necessary to compel respect for the statutory protections Congress has placed around testimony in federal courts. Exclusion is also necessary to remove the incentive to disregard the statute. The benefits of deterrence outweigh the evil of excluding relevant evidence, and the balance falls heavily in favor of suppression.
The panel further explained, id. at 1360, that:
Congress evidenced an intent in [the anti-gratuity statute] to remove the temptation inherent in a witness’s accepting value from a party for his testimony. That temptation, even if unconscious, is to color or falsify one’s testimony in favor of the donor. The law already imposes on every witness the solemn and fundamental duty to testify truthfully, and accepting unlawful gratuities or inducements from a party compromises that solemn duty. When testimony tainted in this way is presented to the courts of the United States, judicial integrity is directly impugned in a way it is not by tangible evidence whose reliability is unaffected by an underlying illegality.
As one might imagine, the reaction to the Singleton decision was swift and varied. Organizations such as the National Association of Criminal Defense Lawyers and other defense organizations applauded the decision as obvious. The Justice Department and National Association of District Attorneys were highly critical of Singleton. Within a week, the U.S. Court of Appeals for the Tenth Circuit vacated the panel’s decision and agreed to hear the matter en banc. The government filed a supplemental brief calling the ruling a “radical departure from history, practice, and established law” and said that it would “make a criminal out of nearly every federal prosecutor” and any federal judge who had approved a plea agreement making a concession for testimony or who had granted a sentencing reduction based in part on a defendant’s truthful testimony.
The en banc panel reversed and held that the anti-gratuity statute did not apply to government lawyers acting within the scope of their official duties. United States v. Singleton, 165 F. 3d 1297 (10th Cir 1999). The en banc court held that “granting lenience in exchange for testimony has created a vested sovereign prerogative in the [United States] government.” Thus, a U.S. prosecutor’s power to grant lenience to a cooperator had “acquired stature akin to the special privilege of kings.” Id., 1301. The Supreme Court denied certiorari.
Accomplice testimony has been a pernicious issue to the fair administration of justice since biblical times. There is too much empirical data of the danger of wrongful convictions owing to perjured or false accomplice testimony to simply blink away these very real concerns. Therefore, consideration should be given to mechanisms to ameliorate the risk of convictions resting on the false testimony of accomplice witnesses.
One suggestion might include directing prosecutors not to share evidence in a case with a cooperating witness. It should be intuitive, but a prosecutor should not share details of tangible evidence or other witness testimony with a cooperating witness. By sharing evidence with the cooperator, the prosecutor may wittingly or unwittingly provide corroborating detail or otherwise signal testimony to the cooperator. Such a prohibition would, of course, be consistent with the prosecutor’s paramount duty to not knowingly assist a witness to testify falsely.
Another suggestion is to require gating procedures for cooperator testimony. The only other circumstance that springs to mind where a witness is paid for his or her testimony is in the case of expert witnesses. There, the law provides for a Daubert hearing to, among other things, allow the district court to screen powerful and potentially persuasive testimony before the jury hears it. A similar procedure can be implemented for cooperator testimony where the testimony can be previewed and cross examined and a determination made about its reliability before it is presented to the jury. As an illustrative example, the state of Illinois requires reliability hearings in capital cases where the would-be cooperator is a jailhouse informant, and it is not unheard-of in state courts to require reliability hearings in certain cooperating-witness cases.
A third suggestion springs from cases such as Hoffa that hold that cross-examination of a cooperating witness and instructions to the jury are sufficient to safeguard the fair trial guarantee. Courts should consider the effectiveness of both in cases of cooperating witnesses. For example, the Jencks Act requires disclosure of prior statements of a witness only after the witness has testified on direct. Oftentimes, as a practical matter, prosecutors will make such information available a week or several days before trial to avoid delays in the conduct of the trial. But, to make meaningful use of prior statements, such statements must be disclosed much earlier. No meaningful investigation of details of prior statements can be made in the hurly burly of final trial preparation. The same is true of Brady and Giglio information. In the case of cooperating witnesses, such information should be disclosed early in the case so that a meaningful and effective cross examination can be conducted.
Additionally, because cross-examination is considered the safeguard against the possibility of perjured or fabricated co-conspirator testimony, substantial latitude should be given to defense counsel at trial to develop evidence of bias and motive and latitude should be given to develop other impeachment evidence, including other acts evidence under FRE 608 and other evidence that impacts on credibility, including Rule 609 evidence of prior criminal conduct.
Similarly, the standard jury instruction applicable to cooperating witnesses should be reexamined. Sand’s Modern Federal Jury Instructions urges the jury to scrutinize accomplice testimony with “great care” and “particular caution.” Some jurisdictions’ instructions are more pointed and emphasize that a cooperator’s testimony should be viewed with “suspicion” or “distrust.” For example, the standard instruction in Illinois says that, “When a witness says he was involved in the commission of a crime with the defendant, the testimony of that witness is subject to suspicion and should be considered by you with caution.”
There are, no doubt, other safeguards that can be considered. What is obvious is that one wrongful conviction based on perjured conspirator testimony is too many and robust safeguards should be in place.
* * *
Author’s note: Benjamin Daniels assisted in the preparation of this article.