On January 13, a federal judge in Connecticut denied Carvaughn Johnson’s petition for habeas corpus relief. In the ordinary course, a habeas denial is a quiet event—a docket entry, an order, a case closed. But Judge Kari A. Dooley did something unusual. After denying Johnson’s petition, s he wrote a separate order directing the State to send the entire record to the Conviction Integrity Unit at the Chief State’s Attorney’s Office. Her reason: the “significant similarities” between Johnson’s prosecution and the wave of civil rights lawsuits now pending against the City of New Haven and its police department, each alleging wrongful convictions produced by officer misconduct during the same era. That order deserves public attention, because it reveals a truth that is otherwise obscure: a federal court can deny relief to a petitioner it may believe might be innocent. Johnson was convicted of murder following a 2004 retrial in New Haven. The only alleged eyewitness to the killing was Ralph Ford. At trial, Ford recanted his prior inculpatory statements and testimony in their entirety. He described being subjected to pressure by New Haven police detectives —allegations that, as Judge Dooley observed, are “strikingly similar” to those raised in at least six pending federal civil rights complaints. Those cases —that of Vernon Horn, Marquis Jackson, Ronald Taylor, George Gould, Stefon Morant, and Adam Carmon —each involve New Haven residents who spent years or decades in prison before their convictions were expunged or they were exonerated, and each alleges misconduct by the New Haven Police Department during the mid-1990s to early 2000s. Judge Dooley further noted “at least some overlap between the detectives involved in Mr. Johnson’s prosecution” and those named in the other cases. (I represent Maceo “Troy” Streater in yet another lawsuit with similar allegations from the same time period.). Despite these observations, the court denied Johnson’s habeas petition. That outcome is not a contradiction. It is a consequence of how federal habeas law works after the Antiterrorism and Effective Death Penalty Act of 1996. AEDPA does not ask whether a conviction is reliable. It asks whether the state court’s prior ruling on each claim was “contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States.” That is a double layer of deference: the federal court defers to the state court, and the state court’s decision is measured only against the narrow body of holdings the Supreme Court has directly addressed. If the Supreme Court has not squarely decided the precise constitutional question presented, the federal court is often powerless to grant relief—regardless of what the record shows. In Johnson’s case, that framework blocked review of claims about a hostile courtroom environment, limits on cross-examination of the sole eyewitness, police failures to document and preserve evidence, and trial counsel’s decision not to present alibi witnesses. Each of these issues would trouble any fair-minded reader of the record. Under AEDPA, none of them cleared the bar. For the public, the critical point is this: “federal habeas denied” is not a judicial finding that a conviction is sound. It is a statement that the petitioner failed to thread a needle that Congress designed to be nearly impossible to thread. A federal judge can believe a case is deeply troubling and still lack the statutory authority to grant relief. Judge Dooley’s extraordinary separate order —directing materials to the Conviction Integrity Unit and citing the pattern of New Haven-era civil rights cases —signals exactly that kind of concern. Judge Dooley is not the first judge to find the Johnson conviction troubling enough to act on. In 2013, Judge Susan Cobb of the Superior Court granted Johnson’s state habeas petition, ruling that he had been deprived of the effective assistance of counsel guaranteed by the United States Constitution and that the deficiency deprived him of a fair trial. Judge Cobb’s decision was a merits determination by a jurist who reviewed the full record and concluded that the constitutional floor had not been met. The Division of Criminal Justice appealed. The Appellate Court reversed, and the Connecticut Supreme Court affirmed that reversal in Johnson v. Commissioner of Correction, 330 Conn. 520. But the fact that a habeas court found Johnson’s trial constitutionally deficient —and that the state had to fight through appellate review to preserve the conviction— should not be forgotten. It means that two judges, over a span of 13 years, have independently examined this case and found it problematic. One overturned the conviction outright. The other, constrained by AEDPA, could not—but took the extraordinary step of directing the record to the state’s conviction integrity apparatus. The question now is whether Connecticut will act on it. Connecticut’s Conviction Integrity Unit, housed within the Division of Criminal Justice, was created to review cases alleging actual innocence or integrity failures. Its published process includes investigation and referral to a Conviction Review Panel that includes a retired judge and a defense attorney. On paper, this is a serious mechanism. In practice, the record is less encouraging. On its website, it has released five reports on concerning cases in the approximately five years since its inception. Investigative reporting by Connecticut Public has documented that the unit has moved slowly and that many cases were closed without an overturned conviction. The Valentine case — its first report — illustrates the problem. There, the Conviction Review Panel met multiple times, requested specific investigative steps, reviewed recordings and files, and then voted that “no action was warranted” —reasoning that no new relevant evidence had been presented beyond what juries and courts had already considered. The panel stated it was not within its “purview to question determinations already made by juries and courts.” That framing converts the CIU into a rubber stamp. If the unit’s operating premise is that it cannot revisit what juries decided, then it cannot serve its stated purpose of catching the cases where juries got it wrong because the evidence was corrupted before it ever reached them. This matters urgently because New Haven’s wrongful conviction problem is not a matter of isolated cases. A February 2025 CT Mirror report documented that eight wrongful conviction compensation claims totaling $37.6 million were pending before the legislature—and that New Haven accounts for a disproportionate share of the state’s wrongful conviction payouts. The cases share recurring features: witness identifications under conditions that modern science identifies as high-risk, allegations that witnesses were fed information, incomplete or missing recordings of police interrogations, and belated disclosure of exculpatory material. The science on eyewitness evidence reinforces the urgency. A 2014 National Research Council consensus report concluded that eyewitness identification accuracy is highly susceptible to contamination absent standardized safeguards —double-blind administration, confidence statements, full recording of witness interactions— that were not standard practice in the era of Johnson’s prosecution. The Innocence Project’s data shows that eyewitness misidentification contributed to roughly 69 percent of DNA exonerations nationally. Johnson’s case is eyewitness-dependent, and the sole eyewitness recanted. The detectives involved overlap with those named in other civil rights lawsuits. The case was tried in an era whose investigative practices the NHPD itself has implicitly acknowledged were inadequate, pointing to modern reforms as evidence of improvement. This establishes that his conviction belongs to a category of cases that demand serious, proactive review —not the passive, excessively deferential process that has characterized the CIU’s work to date. Here is what should happen. The Conviction Integrity Unit should treat Judge Dooley’s order as what it is: a federal judge identifying a pattern of concern and directing materials for review. The CIU should conduct a genuine investigation of the Johnson prosecution, unencumbered by the premise that prior jury verdicts are presumptively correct when the integrity of the evidence underlying those verdicts is precisely what is in question. More broadly, the CIU—or an independent body with access to CIU resources—should audit all New Haven-era contested innocence cases sharing the risk factors that recur across the pending civil rights suits: witness coercion allegations, identification irregularities, missing recordings, late-disclosed evidence, heavy reliance on a single witness, and recantations. A federal judge looked at this case and saw a pattern. Another judge, over a decade ago, found he never got a fair trial. Connecticut’s institutions should be willing to see how troubling this is and do something about it. Carvaughn Johnson has been incarcerated for over two decades for a murder conviction that rests on testimony its only eyewitness has repudiated, obtained through methods that multiple federal lawsuits allege were corrupt, by detectives whose conduct in other investigations has led to exonerations and millions of dollars in compensation claims. The procedural barriers of federal habeas law prevented a federal court from granting him relief. Those barriers do not exist in Hartford. The question is whether anyone there is willing to act. Alexander Taubes is an attorney in New Haven. ...read more read less