Opinion: A prison reform law means nothing if no one can enforce it

On Monday, the Supreme Court will hear argument in a case brought by a man most of the country has never heard of, over a sentence credit most Americans could not describe, in a statute most members of Congress who voted for it have likely forgotten the details of. Maxwell v. Dinis sounds like the kind of case that only lawyers could love. It is actually a test of whether one of the more genuinely bipartisan laws of the last decade means what it says, or whether it is merely a set of instructions the Bureau of Prisons can ignore without consequence.

The law is the First Step Act, signed by President Trump in December 2018 after passing the Senate 87-12 and the House 358-36 — a level of agreement that, in the Washington of 2026, now reads like a dispatch from a different country. Among its provisions, codified in the federal criminal code, is a simple exchange: inmates who complete recidivism-reduction programming earn "time credits" that must be applied toward earlier transfer to a halfway house, home confinement or supervised release. The statute's text uses the word "shall," not "may," when it describes the Bureau's obligation to apply those credits and move eligible prisoners out of secure custody.

A Promise Written Into Statute, Ignored in Practice

The trouble is that "shall" has not meant much in practice. According to an amicus brief filed in August by Rep. Jamie Raskin, the ranking Democrat on the House Judiciary Committee, the Bureau of Prisons has fully applied earned credits to only 29 percent of eligible prisoners, and more than half of those eligible would already qualify for transfer out of prison had their credits simply been applied as the statute requires. The Bureau's own description of the program, laid out on its First Step Act resource pages, promises a system of earned credit toward earlier release; the gap between that promise and the Bureau's own implementation record is the quiet scandal sitting underneath this case.

William Maxwell is one data point in that gap. He is serving a 20-year racketeering sentence at FCI Beaumont Low in Texas and, representing himself, filed a petition arguing the Bureau had wrongly withheld time credits that would have moved him into a halfway house or home confinement. The Fifth Circuit threw the petition out — not on the merits of whether he had earned the credits, but on the theory that a habeas corpus petition under 28 U.S.C. § 2241 is the wrong vehicle entirely unless winning would automatically shorten a sentence, as opposed to merely changing the conditions of custody. It is the kind of procedural trapdoor that lawyers recognize instantly and that non-lawyers rightly find bewildering: the court did not say Maxwell was wrong about the credits. It said he was not allowed to ask.

A Rare Case the Government Itself Says Was Wrongly Decided

What makes Maxwell's case unusual is not just that the Supreme Court agreed to hear it — a vanishingly rare outcome for a petition originally filed pro se from a prison cell, before outside lawyers stepped in. It is that the government does not even defend the ruling Maxwell is appealing. In its own brief to the Court, the Solicitor General's office said the Fifth Circuit's reasoning was inconsistent with the Supreme Court's own habeas precedent, while still urging the justices not to take the case — a position so awkward that the Court had to appoint an outside lawyer simply to argue in defense of the judgment nobody at the table actually wants to defend. That is a telling posture for the executive branch to be in: the administration agrees the Bureau of Prisons it oversees may have gotten the law wrong, yet it would rather the question not be settled.

"Congress made a promise. The courts must make sure the government keeps it."

That line, from Raskin's amicus brief, is about as close to a thesis statement as this case has. Senators Dick Durbin and Cory Booker made a parallel argument in their own amicus filing, telling the justices that nothing in the statute's text, its legislative history or existing precedent supports reading the First Step Act as unenforceable. Maxwell's lawyers point to nine other circuit courts that have allowed similar time-credit claims to proceed through habeas, against only the Fifth and Eighth Circuits' narrower view — the kind of split the Supreme Court exists to resolve.

The Counterargument Deserves to Be Taken Seriously

It would be too simple to say the government's own lawyers are the only ones with a real argument here. The second question the Court agreed to decide — whether Maxwell even exhausted his administrative grievances before going to court — is not a throwaway technicality. Habeas corpus has historically been reserved for claims that a person is being held unlawfully, not a general-purpose tool for managing the day-to-day administrative decisions of a prison system that holds more than 150,000 people. If every disputed program credit, custody classification or disciplinary sanction could trigger a federal habeas case, district courts could be swamped, and Congress did build an internal grievance process into the Bureau of Prisons for exactly this kind of dispute. Judges are right to be cautious about turning habeas into an all-purpose appeals channel for prison administration, and even the government, while backing Maxwell's reading of the statute, has argued his case is a poor vehicle for settling the broader question.

But caution about docket management is a reason to police exhaustion requirements carefully — as the second question presented in this case already does — not a reason to let the Bureau's compliance with a federal statute escape judicial review altogether. Those are different problems, and the Fifth Circuit's rule does not distinguish between them. It forecloses the courthouse door regardless of whether an inmate exhausted every internal remedy available, so long as the relief sought is a transfer rather than an outright release. A grievance process is only a meaningful alternative to a lawsuit if the agency actually has to answer for ignoring it, and a 29 percent full-compliance rate is not evidence that internal grievances are doing that job.

What a Technical Habeas Ruling Actually Decides

It is worth saying plainly what is really being litigated here. Congress built an incentive system explicitly meant to reduce reoffending by rewarding participation in drug treatment, job training and other programs — the kind of policy that earns support from law-and-order conservatives and criminal-justice reformers alike, which is exactly why it passed with such lopsided majorities in 2018. A statute that depends entirely on an agency's voluntary good faith for its enforcement is not really a statute; it is a request. If the Bureau of Prisons can simply decline to apply credits it owes, and the people those credits belong to have no court to appeal to, then the "shall" in 18 U.S.C. § 3624 is doing no legal work at all.

The Supreme Court does not need to turn every prison grievance into a federal case to rule for Maxwell. It needs only to hold that when Congress tells an agency it "shall" release earned credits toward a prisoner's transfer, and the agency does not, the person denied that benefit has somewhere to say so. That is a modest proposition. It is also, judging by how rarely the Bureau has honored its own obligations under this law, an urgently necessary one.