Beyond Formal Dissent: How to Prevent the Next Wrongful Conviction

NASA Learned from Challenger – and Columbia Still Happened

In my previous post, I suggested that NASA’s Formal Dissent Process might offer a useful model for criminal justice. The response I received, and a closer look at NASA’s own history, led me to a harder conclusion: formal dissent is necessary, but not sufficient.

The reason is Columbia. After the Challenger disaster, NASA undertook extensive reforms intended to improve internal communication, oversight, and the handling of dissent. Yet seventeen years later, Columbia was lost in another disaster marked by strikingly similar organizational failures.

The deeper challenge, then, is not simply to create a formal channel for dissent, but to build an organizational culture in which dissent is actively valued, challenging information cannot be filtered out before it reaches decision-makers, and failures, near misses, and “good catches” become routine opportunities for collective learning.

The problem is hardly abstract. On August 17, 2026, Gothamist published a disturbing investigation into one of New York’s most serious wrongful-conviction cases, arising from the Queens District Attorney’s Office. Three men – George Bell, Rohan Bolt, and Gary Johnson – spent a combined nearly 75 years in prison for murders they did not commit. When their convictions were vacated in 2021, Judge Joseph Zayas found that prosecutors had possessed police reports pointing to other suspects, as well as significant information concerning the prosecution’s key witness, none of which had been disclosed to the defense. He wrote that the prosecution had “completely abdicated its truth-seeking role.”

Following the ruling, disciplinary complaints were filed against seven prosecutors. Five years later, the attorney grievance committee concluded that there was insufficient evidence that the prosecutors had violated the rules governing attorney conduct and closed the complaints. Its notice consisted of only five sentences and did not explain what had been examined or how the decision had been reached.

James Doyle published an essay prompted by the case. He argues that we should not stop at the obvious question: Which prosecutor was at fault? When a failure this serious occurs, we must also examine the organizational system in which the individuals were operating. Drawing on Diane Vaughan’s analysis of the Challenger disaster, he asks whether failure sometimes results not from deviation from organizational norms but from conformity to them – from work processes and professional cultures that gradually normalize dangerous practices.

The Bell-Bolt-Johnson case in Queens is not an isolated example. In a July 2026 overview, the Death Penalty Information Center reported more than 600 instances in which a capital conviction or death sentence had been overturned, or a death-row prisoner exonerated, as a result of prosecutorial misconduct. Of the 202 death-row exonerations then identified since 1973, at least 144 – roughly 70 percent – involved prosecutorial misconduct. Withholding favorable evidence was the most common category.

In recent years, several people who were junior members of the justice system while major criminal cases were making headlines have spoken to me candidly, years later, about what they witnessed and thought at the time. In one case, which ended in a unanimous full acquittal after nearly five years, the deputy prosecutor on the case told me years later that she regretted not withdrawing from it in real time. An intern in the same prosecution office told me that he had protested and asked how an indictment could possibly be filed on the basis of the evidence before the team.

In another case, involving an appeal against a murder conviction before the Supreme Court, a former intern told me that the interns were convinced it was a wrongful conviction. They even discussed resigning together as an act of protest, but ultimately did not do so. The appeal was dismissed, and thirty years later she – along with many others – remains convinced that an innocent person was sentenced to life imprisonment.

I do not recount these stories to blame them. Quite the opposite. I appreciate the candor it took to tell me about them in hindsight. I would also like to think that, in their place, I would have insisted more forcefully – but I do not know.

What is troubling in both cases is that the doubt was already inside the system. People who had seen the evidence believed that something was wrong. The problem was not simply a lack of information, but the inability of professional doubt to alter the course of decision-making.

In such a situation, it is not enough to tell employees, “Be brave.” When the cost of challenging a superior is immediate and tangible, while the prospect of changing the decision is uncertain, an organization cannot rely on personal heroism. It must give professional doubt a pathway, protection, and institutional standing.

NASA offers both a model and a warning. Seventeen years after Challenger, Columbia was lost in another disaster displaying strikingly similar organizational patterns. The Columbia Accident Investigation Board concluded that systemic causes identified after Challenger had not been adequately corrected. Vaughan, whose classic study analyzed the Challenger launch decision, later participated in the Columbia investigation and wrote about encountering many of the same organizational patterns again.

The lesson is crucial: formal safeguards cannot guarantee safety. Their value depends on whether they become part of a durable safety culture rather than another layer of procedure.

NASA’s current project-management framework nevertheless shows what a formal channel for dissent can look like. Its Formal Dissent Process requires full and open discussion, with diverse views fostered and respected in an environment of integrity and trust, without suppression or retribution. Significant unresolved disagreements must be elevated to higher levels, and the process is embedded within a broader system of checks and balances.

For criminal justice, that suggests a formal right to raise professional dissent, documentation of the disagreement, escalation beyond the immediate supervisor, a reasoned response, a right of appeal, and protection against retaliation.

But protection is only the beginning.

One of Doyle’s most important observations is that the internal critic should not merely be shielded from punishment. The organization should positively value that role. No one becomes a prosecutor in order to convict the wrong person while leaving the actual offender free to find more victims. Responsible skepticism can therefore be understood not as disloyalty to the prosecution team, but as loyalty to its deepest professional purpose.

That has practical implications at every rank and career stage. Junior prosecutors and interns should be told from the beginning that thoughtful critique is part of good professional performance. Senior prosecutors should be trained not merely to tolerate skepticism, but to solicit it, recognize its value, and respond productively. Participation in careful internal critique can become professionally meaningful, educational, and even a source of pride.

There is a legitimate concern that such procedures could become a recipe for paralysis: endless objections by junior employees who do not bear responsibility for difficult borderline decisions. Doyle’s response is persuasive. The answer is not to suppress dissent, but to teach people how to dissent well – and managers how to receive it. Training in productive critique should begin early, while leaders should learn to invite skepticism and distinguish responsible challenge from obstruction.

Formal review has another vulnerability. Multiple downstream checks can still fail if challenging information is filtered out upstream. The Queens case illustrates the problem starkly: a reviewer cannot assess exculpatory or contradictory evidence that never reaches the reviewer. A chain of approvals is not a true safety redundancy if every level sees only the case as framed by the same original decision-makers. Meaningful review therefore requires access to material that weakens the prevailing theory, unresolved investigative alternatives, and documented professional dissent.

The sources of challenge should also extend beyond the lawyers handling the case. Police investigators, forensic laboratory personnel, academic specialists, victims’ rights advocates, and others may see risks that prosecutors do not. In appropriate cases, a designated devil’s advocate – even a prosecutor formally assigned to articulate how the defense might understand the evidence – could provide another line of protection against collective tunnel vision.

Doyle also proposes learning systematically from what medicine calls “good catches”: situations in which someone notices a danger in time and prevents harm. Criminal justice should do the same. If an intern, investigator, prosecutor, defense lawyer, judge, forensic scientist, or other participant notices a danger and helps avert a wrongful outcome, the event should become an object of organizational learning. The institution should identify what allowed the warning to be recognized, what nearly prevented it from being heard, and how those conditions can be reproduced. The internal critic should be treated as a safety resource, not as a problem to be managed.

This is especially important for younger members of the system. Doyle adds an important dimension to the familiar language of “all stakeholders” and “all ranks”: all ages. The interns in the cases I described were not peripheral to the system. They were part of its informational environment. A genuine safety culture should therefore mobilize people across ranks, roles, ages, and levels of experience.

The same logic applies to retrospective learning. The Quattrone Center’s review of the Lex Street Massacre offers a concrete criminal-justice example. A sentinel-event review brought multiple stakeholders together to examine how four innocent men came to be wrongfully incarcerated for 18 months.

The significance of that model lies not only in broad participation, but in the nature of the process. There is an important difference between an assessment conducted to the people involved and one conducted with them. A top-down panel of experts can produce valuable findings, but a collaborative assessment involving all relevant stakeholders can reveal interactions that no single institution can see from within its own silo.

Doyle notes another advantage of this collaborative model: the Quattrone projects produced unusually broad consensus around factual findings and reform recommendations. Because those recommendations were products of broad stakeholder teams, they were more likely to be accepted and implemented.

The dissent process itself must also be subject to scrutiny. After any significant safety event, the organization should examine not only what happened in the case but how its warning mechanisms performed: whether concerns were raised, documented, and escalated, how supervisors responded, and whether the process had the capacity to change the outcome. The evaluation process must itself be evaluated.

Regularly examining how the mechanism operates can also demonstrate that internal criticism is structured, conscientious, and directed toward better decision-making.

In a prosecution service, these ideas could be combined into a broader Safety Dissent Process. Any intern, investigator, or prosecutor who identifies a substantial risk of wrongful conviction, disregarding exculpatory evidence, or another serious failure should be able to initiate a documented review outside the original case team, receive a reasoned response, escalate the matter if necessary, and be protected against retaliation. Crucially, responsible criticism should be positively valued, and reviewers should have independent access to information that challenges the prevailing theory.

Not everyone who raises a doubt will be right. Nor should a dissenter have a veto. The purpose is more modest – and more important: to ensure that serious professional doubt does not disappear merely because it was voiced by the most junior person in the room.

This is also the approach we are trying to advance at the Institute for Safety in the Criminal Justice System. A safe system is not one in which no one ever makes a mistake. It is a system capable of identifying warning signs in time, listening to those who notice them, learning systematically from safety events, and reconsidering decisions before harm becomes irreversible. The Institute’s latest report, covering 2024-2025, analyzes eight new safety events and conducts a cross-case analysis of the eight events examined in the previous report to identify recurring patterns and propose safety rules. It also emphasizes the aspiration to establish a National Institute for Safety in the Criminal Justice System with the resources and authority needed to implement safety mechanisms systematically.

The lesson from NASA is therefore not that a Formal Dissent Process will prevent the next disaster. NASA’s own history rules out so simple a conclusion. The deeper lesson is that safety requires institutions to make dissent visible, protect and value those who raise it, prevent challenging information from being filtered out, broaden the sources of criticism, and learn systematically from failures, near misses, and good catches.

When someone raises a hand and says, “Something here is not right,” a safe system cannot merely permit that person to speak. It must become an organization that wants to hear them.

I am deeply grateful to James Doyle for his generous comments on an earlier draft. Several of the ideas developed in this essay – including positively valuing internal critics, all-ages participation, training in productive dissent, good catches, collaborative multi-stakeholder reviews, and reviewing the dissent process itself – directly reflect his suggestions.