Reversing the Malcolm X Convictions: How It Happened, How Far We’ve Come, How Far We Need To Go
Since 2014, when Ken Thompson became Brooklyn district attorney after making wrongful convictions the centerpiece of his campaign, New York City DAs have voluntarily overturned more than 50 homicide, robbery, and rape convictions. An extensive body of knowledge about the causes of wrongful convictions, built by lawyers and experts over the past three decades, has provided the basis for these actions. Late this year saw the most highly publicized reversal of them all: Manhattan DA Cyrus Vance’s overturning of the 55-year-old murder convictions of two of the alleged assassins of Malcolm X.
The Malcolm X case has two of the features that pervade wrongful conviction cases: faulty identification procedures and police and prosecutorial misconduct. This case is fascinating to analyze, not only to expose what was done wrong 55 years ago, but also as a basis for showing how far we have come. The procedural safeguards employed back then were primitive compared with what we have now. And yet, a conviction as wrongful as this one could happen again. We have come far, but not far enough.
It was 3 p.m. on February 21, 1965, at the Audubon Ballroom in upper Manhattan, when the brilliant and charismatic Black militant author and orator, Malcolm X, began to speak. “Salaam Alaikum,” he said, “peace be unto you,” as he greeted his audience of 400 Black followers (whites having been excluded) – which included police undercover agents, FBI informants, and assassins.
Malcolm, less than a year earlier, had broken with the Nation of Islam’s militant separatist leader, Elijah Muhammad, over both personal and political differences. No longer rejecting the possibility of coexisting with whites of good will, Malcolm now led a new organization, the Organization of Afro-American Unity, and this was its event.
Malcolm had been the subject of death threats and had told journalists and friends he expected to be assassinated. Just a week earlier, his home in Queens had been firebombed and he and his family were living at a midtown Manhattan hotel. Another speaker told the audience that Malcolm was a man “who would lay down his life for you.” Now, Malcolm was beginning to speak.
Individuals in the crowd yelled about a fictitious pickpocketing attempt, creating a diversion, and someone threw a smoke bomb into the crowd. Just then, a man wielding a sawed-off shotgun stepped forward and shot Malcolm, knocking him over. At least two other men, armed with a .45 caliber semi-automatic pistol and a 9 mm Luger, rose from the audience and fired continuously into Malcolm’s prone body, hitting him at least 15 times. During the bedlam, Malcolm’s pregnant wife, Betty Shabazz, could be heard sobbing, “they’re killing my husband,” as she threw herself on top of her daughters to protect them from the fusillade.
One of the killers, Mujahid Abdul Halim, was shot in the leg by Malcolm’s bodyguards and apprehended at the scene, but the other shooters and conspirators escaped. Within a couple weeks, after an extensive FBI and NYPD investigation, the two additional alleged shooters had been apprehended and charged. They were Muhammad A. Aziz, also known as Norman 3X Butler, and Khalil Islam, also known as Thomas 15X Johnson, both members of the Nation of Islam.
At the three defendants’ joint trial one year later in the Supreme Court, New York County, 10 eyewitnesses identified Halim, the man arrested at the scene. His conviction proved easy: he testified at the trial and confessed his guilt. Seven of the same eyewitnesses identified Aziz, Islam, or both. Several knew one or both of them or claimed to have had excellent opportunities to observe the shooters. Aziz and Islam were convicted too.
Many who followed the trial and studied it later harbored gnawing doubts. Both men had testified they were elsewhere when the crime occurred. They were supported not only by alibi witnesses but by Halim’s own testimony exonerating them and blaming others he refused (at that time) to name. No physical or forensic evidence implicated the defendants or placed them at the crime scene, the eyewitness accounts differed in significant respects, and a key prosecution witness who knew the defendants beforehand was shown to have falsely denied a serious psychiatric history.
Meanwhile, although Malcolm’s death seemed to serve the interests of Elijah Muhammad and his top aide, Louis Farrakhan, neither the FBI nor the NYPD ever made a case against any higher-ups. The only one to act against them was Malcolm’s own daughter, who in 1995 pled guilty to conspiring to murder Farrakhan to avenge her father.
Over the ensuing five decades, many historians, authors and filmmakers concluded that Aziz and Islam were innocent. Halim signed affidavits, used by Aziz and Islam in an unsuccessful 1978 motion for a new trial, naming the true conspirators and describing the plot in great detail. In the early 1990s, Spike Lee released a movie starring Denzel Washington and wrote a book naming other suspects, while pointing a finger at Elijah Muhammad and Louis Farrakhan for, at the very least, inciting the lethal event through their virulent rhetoric. However, no reinvestigation occurred. Aziz and Islam served out their sentences and were paroled.
In January 2020, after a new book and a television documentary again focused on the case, Manhattan DA Vance agreed to reinvestigate their convictions through his office’s Conviction Integrity Program, in cooperation with new attorneys for Aziz and Islam (the latter had died in 2009), David Shanies of the Shanies Law Firm and Barry Scheck of the Innocence Project. The 20-month inquiry resulted in Vance’s appearance in court on November 18, 2021, together with a leader of his Conviction Integrity Program, Charles King, the defense attorneys and Aziz, to jointly move to vacate Aziz’s and his late co-defendant’s 55-year-old convictions.
During the proceeding, Vance apologized to Aziz for the actions of his office. Aziz accepted the apology, but stated, “I am an 83-year-old who was victimized by the criminal justice system. . . . I do not need this court, these prosecutors or a piece of paper to tell me I’m innocent.” He added that his story is “all too familiar to Black people, even in 2021.” What had gone wrong?
According to the parties’ joint motion, the main thing that went wrong was that law enforcement, including the NYPD, the FBI and the DA’s office, failed to disclose evidence favorable to the defense, or Brady material, the timely disclosure of which might well have led to acquittals. About the myriad Brady violations, the joint motion was explicit. Implicit was that investigators had also relied on faulty identification procedures which likely caused the conviction of two innocent men. The written motion was careful to avoid the issue of guilt or innocence, but Vance’s apology and the motion’s tone certainly implied that the wrong men had been convicted and their lives tragically destroyed.
In 1963, just three years before this trial, the Supreme Court had decided Brady v. Maryland. The court held “that the suppression by the prosecution of evidence favorable to an accused upon request violates due process where the evidence is material either to guilt or to punishment, irrespective of the good faith or bad faith of the prosecution.” Not until 1972, six years after the trial, did the court clearly announce that impeachment material also had to be disclosed under the Brady rule and that a prosecutor was strictly liable to disclose all such evidence in the possession of the police or the prosecutor’s entire office. Not until 1976 did the court decide that such material had to be turned over regardless of whether the defense made a specific request. Clearly, when the Malcolm X trial occurred in 1966, the Brady rule had not taken hold in law enforcement. Indeed, the obligation to disclose evidence that might defeat a prosecutor’s case did not take hold for decades, if it ever has, as dozens of subsequent reversals – the tip of the iceberg of Brady violations, most of which remain forever secret – later revealed.
As the joint motion notes, it is an open question whether the post-trial decisions expanding the Brady rule would apply retroactively to a trial that occurred in 1966. Had Aziz and the Islam Estate brought a contested post-judgment motion alleging Brady violations, Vance’s appeals bureau would almost certainly have argued that the relevant Supreme Court decisions were not retroactive. But conviction review units may focus on the integrity of a criminal judgment without hiding behind procedural bars. Here, Vance, knowing of a new witness who had supported Aziz’s alibi defense and that Aziz had passed a polygraph test, decided to waive any retroactivity defense. (One might ask why appeals bureaus in a close case don’t do the same, but that is a question for another article.)
While noting that the FBI had deliberately concealed crucial information from the NYPD and the DA’s office, Vance conceded that the three agencies were “partners” in the prosecution and thus the people were responsible for the contents of all three agencies’ files. Collectively, the FBI and the NYPD had largely withheld from the DA’s office, and thus from the defense, that police and FBI undercover detectives and informants had been present, had witnessed the events, and had provided accounts which conflicted with the prosecution’s theory and supported the defense’s. Indeed, one such NYPD undercover officer had infiltrated Malcolm X’s security team and had provided an exculpatory account. The failure to disclose that trained law enforcement observers were present also deprived both the DA’s office and the defense of the opportunity to interview them and to use their potential testimony.
In addition, FBI and NYPD reports revealed that several witnesses had named third parties as the culprits or had provided descriptions of how the events had occurred, or of the shooters, which favored the defense. The suppression of these reports deprived the defense of the opportunity to investigate these important, potentially exculpatory leads.
Vance’s motion also detailed direct suppression of Brady material by his own office. It unquestionably had in its file NYPD reports that put the shooters in a completely different position than the men identified by the eyewitnesses as the defendants. One report contained an interview with a witness who knew Aziz and Islam well and said he did not see them in the ballroom. The file also contained letters from the eyewitness with the psychiatric issues, which should have been disclosed as well. While Vance could not say with certainty that these reports were in the people’s file at the time of trial, let alone withheld from the defense, his motion essentially assumed this to be the case.
The joint DA/defense motion implicitly criticized the NYPD’s identification procedures in this case where identity was the only real issue. The motion noted trial testimony by one witness who identified Aziz that the lineup he viewed consisted of “men of different sizes and complexions,” with only two men, one of whom was Aziz, wearing a gray coat similar to the gray coat he had previously described, while the other of the two obviously did not fit the rest of his description. The motion notes that the defense challenged other identification procedures as well. However, the defense was at an impossible disadvantage because, in 1965, the Supreme Court had not yet imposed the constitutionally mandated procedures we now take for granted to combat unduly suggestive and unreliable pretrial identification procedures.
United States v. Wade, which established the right to counsel at station house lineup procedures, and its companion case, Stovall v. Denno, which held that unreliably suggestive identifications could be excluded at trial as a matter of due process, were not decided until a year after the Malcolm X trial (and the court refused to make the new rules retroactive). Police thus were free to conduct showups and lineups without any defense lawyer or independent observer watching and without their procedures being reviewed before trial at what has since become known as a “Wade hearing.”
In 1965 (and indeed, for nearly half a century afterward), there were no or limited “best practices” in place to minimize the risk of unfair suggestion. Police rarely documented the procedures they used, and any photographs of the lineup generally were small and blurry. They often made no record, before conducting a lineup, of the details of an eyewitness’s description of the perpetrator. They made no record of what was said to the eyewitness or what the eyewitness said during or after a lineup. A detective could openly or subtly hint to the eyewitness which individual the suspect was, make no record of a witness’s uncertainty, and confirm for the witness the “correctness” of a tentative identification, with virtually no risk of detection. He or she could record all positive identifications while making no record of exculpatory results, such as a statement that “the shooter is not there.” As a result, although the Malcolm X trial turned entirely on eyewitness identifications, the identification procedures that were conducted were not then, and are not now, susceptible to meaningful review. We will never know how many exculpatory witness interviews or lineups were conducted that were not documented.
The joint motion notes that, in addition to the seven witnesses who identified Aziz and/or Islam in court, seven more did so out of court from photographs but did not testify. That Vance agreed to vacate the conviction anyway implies that he lacked confidence in the procedures used and that at least 14 individuals made false identifications. If Vance lacked confidence in a conviction with 14 identification witnesses, what confidence can we have in other convictions obtained during that era, many of which, no doubt, rested on one or a couple identifications? The surviving defendants or their families in other criminal cases from this era should now clamor for their cases to be reconsidered too.
Consider this: the Final Report of the New York State Bar Association’s Task Force on Wrongful Convictions, chaired by former Supreme Court Justice Barry Kamins, and approved by NYSBA’s House of Delegates in April 2009, noted that of 53 reversed convictions it studied where the defendants appeared innocent, 36 involved erroneous identification testimony. An updated Task Force Report, approved by NYSBA’s House of Delegates in April 2019, highlighted that 70% of DNA-based exonerations nationally involved at least one misidentification, and 53% of DNA-based exonerations in New York State involved at least one misidentification. Of the 253 New York State-based exonerations listed by the National Registry of Exonerations, 88 also involved at least one misidentification. Of the 192 cases the Innocence Project has gotten overturned since 1992 based on DNA evidence definitively establishing innocence, 63% involved false identifications. Psychologists have conducted controlled studies, under presumably less stressful circumstances than real-life violence, and established that more than a third of eyewitness identifications are wrong with the error rate more than double that where poor fillers make the lineup suggestive.
Meanwhile, convictions based upon eyewitness identifications have not received searching factual reviews. As the State Bar Wrongful Conviction Committee wrote in 2009, the Appellate Division usually defers to the fact-finding of hearing judges and juries, the Court of Appeals lacks factual review jurisdiction, and federal habeas judges are required to defer in virtually all but the most extreme cases to the factual findings of the state courts.
When one considers the procedures and best practices that are followed today, or at least supposed to be followed, we realize how far our system has come. First, we are no longer so dependent upon police-conducted interrogations and identifications, for a wealth of other evidence is now available to investigators. The assassination of a public figure like Malcolm X in front of 400 people would be recorded now by security cameras and countless smartphones. The perpetrators’ associations, planning, locations and bragging likely would be detected on social media and in their cellphone records. Their DNA might be detected on objects they left behind or handled on the way out.
Second, the methods used by police investigators are better regulated and subjected to more searching judicial review. Three months after the Malcolm X trial, the Supreme Court decided Miranda v. Arizona, requiring criminal suspects in custody to be warned of their constitutional rights. Because police coercion thereafter still led to numerous wrongful convictions, the New York State Legislature, in 2018, following the State Bar Report’s recommendation, required that police video record interrogations of suspects, in murder and certain other serious felony cases. Police are now also taught other best practices, such as making a detailed, verbatim record of both police and witnesses’ statements during identification procedures. Police are required by statute to conduct photo identifications blindly – that is, by a detective who does not know who the suspect is – and the Division of Criminal Justice Services protocol recommends that lineups be conducted blindly as well.
Third, and perhaps most important, the state Legislature in 2020 enacted comprehensive discovery reform, under which the prosecution must disclose, early in the proceedings, virtually everything in its file and the police file. Formerly, state prosecutors had to provide only limited categories of materials enumerated by statute and, just before opening statements, the prior recorded statements of the prosecution’s witnesses under the Rosario rule. Even the most skilled, conscientious defense lawyers – and many of those assigned to represent indigent defendants were not that – had little opportunity to investigate these materials or plan a defense. As in the Malcolm X case, prosecutors often failed to search law enforcement and prosecutorial files not in their personal possession for exculpatory or impeachment evidence, notwithstanding the Supreme Court’s requirement that they do so. Now, however, prosecutors must formally certify they have searched their own and law enforcement files for virtually every imaginable type of impeachment or exculpatory evidence and provide it on a timely basis or else face sanction, including dismissal.
While we have made great progress, the potential for wrongful convictions obviously remains. Single-eyewitness cases still are prosecuted. Some detectives and prosecutors still fail to implement all the constitutional, statutory and administrative rules and protocols meant to keep them honest. Even honest ones often still suffer from tunnel vision, which may cause them to shape witnesses’ accounts and other evidence to fit their theory of guilt. Prosecutors still are taught not to take notes, which causes unrecorded exculpatory or impeachment material to be withheld, “open file” discovery or not. Indeed, suspect interrogations must be recorded, but not interrogations of mere witnesses. All this occurs in a legal system still infected by racial prejudice. As Aziz rightly said upon his exoneration, many of the hallmarks of wrongful convictions are “all too familiar to Black people, even in 2021.”
One would like to think that today, if a similar crime occurred, surveillance video or other physical evidence would protect a Muhammad Aziz and Khalil Islam from wrongful arrest and prosecution or, at the very least, modernized criminal procedure law would give their lawyers the tools they need to present a meritorious defense. But in a system run by humans, mistakes and misconduct will always occur. We must continue to strive to correct the unspeakably tragic errors of the past, while at the same time learning from them so they don’t happen again.
Joel B. Rudin is the principal in a four-attorney criminal defense and plaintiff’s civil rights firm based in Manhattan. He has won three cases before the U.S. Supreme Court, numerous 2nd Circuit decisions upholding the rights of former criminal defendants to recover money damages for police and prosecutorial misconduct, and numerous large settlements. The New York State Association of Criminal Defense Lawyers awarded him its prestigious Thurgood S. Marshall Award for his work freeing the wrongfully convicted. He has testified about the lack of accountability of prosecutors before the NYSBA’s Task Force on Wrongful Convictions and chaired its Subcommittee on Conviction Integrity Units.
 373 U.S. 83 (1963).
 Id. at 87.
 Giglio v. United States, 405 U.S. 150 (1972).
 United States v. Agurs, 427 U.S. 97 (1976).
 Affirmation of Assistant D.A. Charles King dated November 18, 2021, at para. 46.
 388 U.S. 218 (1967).
 388 U.S. 293 (1967).
 Stovall v. Denno, 388 U.S. 293, 300–01 (1967).
 Joint Mot. to Vacate Judgments of Conviction and Dismiss Indictment, People v. Muhammad Aziz and Khalil Islam, Ind. No. 871/1965, 8 n. 9 (Sup. Ct., N.Y. Co. Nov. 18, 2021).
 New York State Bar Association, Final Report of New York State Bar Association’s Task Force on Wrongful Convictions, 45 (Apr. 4, 2009), https://nysba.org/NYSBA/Practice%20Resources/Substantive%20Reports/PDF/FinalWrongfulConvictionsReport.pdf.
 New York State Bar Association, Report of New York State Bar Association’s Task Force on Wrongful Convictions, 85 (Feb. 8, 2019), https://archive.nysba.org/WorkArea/DownloadAsset.aspx?id=92500.
 Explore the Numbers: Innocence Project’s Impact, Innocence Project, https://innocenceproject.org/exonerations-data/, last visited Nov. 29, 2021; Alexis Agathocleous, How Eyewitness Misidentification Can Send Innocent People to Prison, Innocence Project, https://innocenceproject.org/how-eyewitness-misidentification-can-send-innocent-people-to-prison/, last visited Nov. 29, 2021.
 See Gary L. Wells et al, Policy and Procedure Recommendations for the Collection and Preservation of Eyewitness Identification Evidence, Law and Human Behavior, 44(1), 3-36 (2020). Another study demonstrated the shortcomings of stranger identification after roughly one-third of responses turned out to be mistaken identifications. See Ahmed M. Megreya, David White, & Mike Burton, The Other-Race Effect Does Not Rely on Memory: Evidence From a Matching Task, Quarterly Journal of Experimental Psychology, 64(8), 1473-1483 (2011). In 2003, a study demonstrated that participants identified a filler in a target absent array 85% of the time after a 12-second exposure and 45% of the time after a 45-second exposure. See Amina Memon, Lorraine Hope and Ray Bull, Exposure Duration: Effects on Eyewitness Accuracy and Confidence, British Journal of Psychology, 94, 339-354 (2003). In addition, the use of poor fillers in a lineup can lead to rates of false identification as high as 70%. See Steve D. Charman, Gary L. Wells, and Stephen W. Joy, The Dud Effect: Adding Highly Dissimilar Fillers Increases Confidence in Lineup Identifications, Law and Human Behavior, 35(6), 479-500 (2011).
 Final Report of New York State Bar Association’s Task Force on Wrongful Convictions, 48-49.
 384 U.S. 436 (1966).
 Video recording is required for interrogations for certain enumerated offenses, including class A-1 felonies, except those defined in article 220 of the penal law; felony offenses defined in section 130.95 and 130.96 of the penal law; a felony offense defined in article 125 or 130 of such law that is defined as a class B violent felony offense in section 70.02 of the penal law. See CPL § 60.45(3)(a). The violation of the video-recording requirement does not automatically require suppression, but rather is one factor for the court to consider in assessing the voluntariness of the statement. See CPL § 60.45(3)(b). The same is true of CPL § 60.25’s rule requiring double-blind identification procedures.
 See CPL § 60.25(1)(a)(ii).
 See New York State Division of Criminal Justice Services, Identification Procedures: Photo Arrays and Line-ups: Model Policy, Municipal Police Training Council (June 2017), https://www.criminaljustice.ny.gov/crimnet/ojsa/standards/MPTC%20Model%20Policy-Identification%20Procedures%20and%20Forms%20June2017.pdf.
 CPL §§ 245.10(1), 245.20(1)-(3), 245.50(1), 245.80.