As we publish, twelve people in Massachusetts are in their fourth day of deliberations in a murder trial that has held international attention. We take no view whatsoever on how it should end. But the shape of the decision they are making — and the fact that their reasoning will never be recorded, published or reviewable by anyone — is precisely the subject of this research.
Published 1 September 2026 · written while deliberations were ongoing · sources linked at the end
Public argument about a case like this collapses, almost instantly, into one question: did she do it? But in this trial that question was never live. The defence has not disputed that the defendant caused the deaths of her three children. Nothing the jury does will turn on it.
What the jury must decide is something narrower, stranger, and far harder — a question about a mind, not an act. In Massachusetts the test comes from a 1967 decision, Commonwealth v. McHoul: a person is not criminally responsible if, at the time of the act, a mental disease or defect left them without substantial capacity either to appreciate the criminality or wrongfulness of what they were doing, or to conform their conduct to the requirements of the law.
Read that twice, because two ordinary words are doing enormous work. Substantial — not total. A person can retain some capacity and still fail the test. And or — the second limb stands alone. Someone who knew perfectly well that an act was wrong, but lacked substantial capacity to stop themselves, satisfies the standard just as much as someone who never knew at all.
The practical consequence is severe, and it is why a jury can be out for days in a case where the act itself is admitted. The jurors are not asking themselves "has the defence convinced us she was psychotic?" They are asking "has the prosecution removed our reasonable doubt that she was responsible?" A juror who finds the psychiatric evidence genuinely difficult to resolve has not reached a stalemate. Under the law as the judge will have instructed them, unresolved doubt on that question points in one direction.
Reporting indicates the jury has four verdicts available, and each carries a different consequence. They are not variations of one outcome; they are four distinct answers to two separate questions — was she criminally responsible, and if so, what was the quality of the intent?
Requires the prosecution to prove criminal responsibility and the aggravating element the theory rests on — in Massachusetts, deliberate premeditation or extreme atrocity or cruelty. The sentence is fixed by statute: life without the possibility of parole.
Criminal responsibility proved, but without the first-degree element. Also a life sentence — but one that carries eventual parole eligibility. The gap between doors one and two is measured in decades of a human life, and turns on a distinction of mental quality that no physical evidence can settle.
Widely reported as "not guilty by reason of insanity", and widely misunderstood as walking free. It is not an acquittal into the street: it triggers a psychiatric commitment process, with confinement in a secure facility and periodic review, and release only if and when a court is satisfied the person is no longer dangerous. It can last a very long time. It is a different door, not an open one.
Available in law, as it is in every criminal trial, and the outcome nobody is arguing for here.
Reporting puts the jury at roughly seventeen hours of deliberation across three days before today. Public commentary reads long deliberation as dysfunction. It is usually the opposite: it is what a properly instructed jury looks like when the case genuinely sits on a knife-edge — and this one does, structurally.
Consider what these twelve people are being asked to do. They must reconstruct the interior state of another person's mind on a specific day in the past. Their evidence for it is the testimony of expert witnesses who disagreed with each other, filtered through weeks of technical psychiatric language. They must then map that reconstruction onto a legal test containing the words "substantial capacity" — a phrase with no measurable threshold anywhere in medicine. And they must do all of it to the standard of beyond reasonable doubt, in a case saturated with grief that any human being would feel.
There is no forensic test that resolves this. No document settles it. It is the hardest category of decision the justice system produces — and it is why cases like this one are worth studying long after they conclude.
Whenever this jury returns, here is what the world will receive: a word. Perhaps a few words, once per count.
Not the reasoning. Not which expert they found persuasive and why. Not whether they wrestled with the second limb of the test or never reached it. Not which piece of evidence moved a juror who arrived thinking the opposite. Not whether they understood that the burden sat with the prosecution — or, if they did, how they applied it. Jury deliberations in these systems are not recorded, not transcribed, and not published. The reasoning that produced the outcome is, by design, permanently unavailable to the public, to scholars, and to any future defendant hoping to know how the same test has been applied before.
A glass-box judgment is the alternative this research is testing. Given the same record, a machine can be required to publish every finding of fact, cite the evidence supporting it, mark what it could not determine, state the strongest argument against its own conclusion, and show the path from the legal test to the outcome line by line — so that anyone, including a critic, can find the exact step where they think it went wrong. That is not a claim that a machine would decide this case correctly. It is a claim about what can be inspected afterwards.
Whether machine reasoning would be better is precisely the open question — and it is not answered by opinion pieces, including this one. It is answered by running the experiment on cases that are over, where the record is complete and nothing anyone publishes can affect a living defendant's fate. That work is on our docket.
We will not tell you whether the jury was right. We do not know, and neither does anyone else who was not in that courtroom for the full trial. What we will note is the same thing we have noted here: that you received an outcome without a reason, and that this was a design choice rather than a necessity.
And when this case is fully concluded — verdict, sentence, appeals exhausted, the record complete and public — it becomes eligible for the docket like any other. Not before. That is the rule, and the rule is only worth anything when it holds in the cases where breaking it would attract the most attention.
← Back to the research question · The docket · Research note 01 · The case against Judgir