Daily e-Edition
Evening e-Edition
Sign up for email newsletters

Sign up for email newsletters
Daily e-Edition
Evening e-Edition
Trending:
Critics of science often claim a “gotcha” moment when researchers revise findings or change their minds. How, they ask, can we trust people who admit they were wrong?
That criticism misunderstands the scientific method. Conclusions are tested, challenged, and revised when new evidence warrants it. Science earns credibility not by always being right, but by rigorously correcting itself when it is wrong.
Connecticut should also recognize this principle. From Yale and UConn laboratories to Pratt & Whitney, Sikorsky, Electric Boat, and the state’s growing bioscience sector, the state’s strength depends in part on testing assumptions, identifying errors, and improving on what came before.
A July 2026 Hawaiʻi Supreme Court opinion asks whether justice can meet the same standard.
In 1990, Daniel Granillo was convicted of kidnapping and sexual assault and sentenced to 40 years. An FBI expert testified that a hair found in his car shared microscopic characteristics with the complainant’s hair and that fibers on her clothing were consistent with material from the car. Prosecutors used this “physical evidence” to corroborate her account.
Years later, scientific reviews exposed the limitations of these methods. In 2017, the Justice Department told Maui prosecutors that the FBI’s hair testimony had overstated what science could prove: microscopic comparisons could identify similarities but not reliably link a hair to a specific person.
The Hawaiʻi Supreme Court ordered a new trial.
The ruling did not establish Granillo’s innocence. It found that his conviction could not stand without a new trial. The court recognized that judges, lawyers, and experts may act in good faith and still produce an unjust result. No one has to be dishonest for a system to get something profoundly wrong.
Courts often focus on misconduct: Did the prosecutor knowingly present false evidence? Was the expert dishonest? Those questions matter, but another matters more: Can the verdict still be trusted?
The Hawaiʻi Supreme Court held that when new scientific developments invalidate expert testimony, defendants need not prove that prosecutors knowingly presented false testimony or that a retrial would likely end in acquittal. The question is whether there is a reasonable possibility that the invalid testimony contributed to the conviction.
That standard shifts attention from blame to reliability. The court called the hair and fiber testimony the prosecution’s centerpiece and said it gave contested witness testimony a “veneer of scientific certainty.” Jurors may give expert evidence unusual weight because it appears objective.
The implicit message was clear: Even if you doubt the witness, trust the science. But in this case, the science could not support what the jury had been encouraged to believe.
That lesson extends beyond the courtroom. Scientific conclusions change as evidence changes. Adjusting a conclusion in response to better evidence is not a failure of science. It is science working. The failure would be refusing to change.
The COVID-19 pandemic showed how difficult that can be for the public to accept. Recommendations shifted, treatments were tested and discarded, and Anthony Fauci became a target of attacks that treated uncertainty and revision as evidence of corruption.
That is the broader lesson from Granillo. Science reexamined a forensic practice and acknowledged that experts had overstated what it could prove. The question remained whether the legal system would demonstrate similar intellectual honesty. In this instance, it did.
The Hawaiʻi opinion also criticizes the U.S. Supreme Court and asserts that the state constitution need not adhere to a federal constitutional minimum the justices believe is declining. Some of that language may be seen as unnecessary and could risk making the ruling appear partisan.
But the underlying principle matters, including here in Connecticut. Our state Supreme Court has long recognized that the federal Constitution sets a national minimum for individual rights and that the Connecticut Constitution may provide broader protections.
In Granillo’s case, that independence allowed Hawaiʻi to reject the idea that a conviction can stand merely because prosecutors were unaware that the expert testimony was scientifically invalid. The constitutional harm lies in the unreliable verdict, not only in the government’s intentions.
The ruling may have ramifications beyond hair and fiber analysis. For example, bite-mark comparisons, certain fire-investigation techniques, and overstated pattern-matching claims have faced serious scientific scrutiny. Highlighting these issues underscores the need for ongoing review of forensic practices.
This does not mean every scientific disagreement should reopen a conviction. The testimony must be shown to be scientifically invalid, and there must be a reasonable possibility that it contributed to the verdict. Still, some may argue that legal finality is essential for justice; however, the reliability of verdicts should take precedence when scientific error is clear.
But finality cannot be the justice system’s highest value. The passage of time does not transform an unreliable verdict into a reliable one. Nor does evidence remain true simply because it was once accepted in a courtroom.
Scientific review corrected the record in Granillo. The Hawaiʻi Supreme Court corrected the judgment.
Every institution makes mistakes, but trust depends on what happens next. Science earns trust by testing itself, exposing errors, and changing when the evidence demands it.
Justice systems must strive to do the same, ensuring that verdicts are continuously evaluated in light of evolving scientific knowledge.
Justice must do the same.
Paul Oestreicher is a strategic communications and policy advisor. He authored “Camelot, Inc.: Leadership and Management Insights from King Arthur and the Round Table.”
Copyright 2026 Hartford Courant. All rights reserved. The use of any content on this website for the purpose of training artificial intelligence systems, algorithms, machine learning models, text and data mining, or similar use is strictly prohibited without explicit written consent.