Inside the Supreme Court’s little-known revision process

With today marking the first day of arguments for the Supreme Court, most would think that the justices’ work from the previous term is over.

But in fact, the justices spend years reworking their opinions after they are initially released, purging them of grammatical, spelling, stylistic, and even factual errors. The court’s decisions take effect immediately, but the opinions—the written rationales behind the decisions— don’t become official until they are published in United States Reports, the official publication of Supreme Court rulings.

The court is required to publish the hulking volumes in print, but there’s no specific deadline. The most recent volume was published in June, five years after the rulings were handed down.

Even so, the justices still have a way of sneaking in more changes after the final printing. The Office of the Solicitor General invoked this obscure process this year to suggest that the court remove a 13-year-old error from its 2003 ruling in Demore v. Kim, a case about detention times for illegal immigrants.

In an unusual move, Acting Solicitor General Ian Gershengorn sent an apology letter to the court last month, admitting that the office submitted inaccurate statistics in Demore v. Kim. The Aug. 26 letter said the office underestimated the average time that illegal immigrants spent in detention while waiting to appeal their case.

The solicitor general’s office had told the court in 2003 that the average detention time was four months. In the letter last month, Gershengorn said average detention time was actually more than a year. The mistake, legal experts say, could have major implications for Jennings v. Rodriguez, a similar immigration case the court is scheduled to hear this term.

Immigration reform advocates have long questioned whether the original four-month estimate was an honest mistake, or whether the solicitor general’s office at the time knowingly misled the court.

“If you look carefully at the letter, it’s clear they actually had a chart back then in 2003 that told part of the story of what was wrong with the numbers,” said Nancy Morawetz, a professor at the New York University School of Law who specializes in immigration rights. “They knew at the time the decision came out that the court was misunderstanding the length of detention.”

Some supporters of mandatory detention think the critical response to the solicitor general’s letter was strategic, not substantive.

Jessica Vaughan, the director of policy studies at the Center for Immigration Studies, says the government is only trying to submit relevant information to the court, showing how mandatory detention is necessary to preserving an immigration system with integrity. “It’s more of a distraction than something meaningful to the case,” she said.

The controversy highlights the solicitor general’s ability to submit unvetted evidence directly to the court, free from the discovery process in lower courts, where both parties have equal footing. Critics have contended for years that the relationship has resulted in systematic submissions of erroneous information.

The solicitor general is appointed by the president to represent the administration before the Supreme Court. It’s a prestigious position, sometimes used as a stepping stone to future bench appointments. Five solicitors general have gone on to the Supreme Court bench, including Justice Elena Kagan, who served as solicitor general from 2009 to 2010.

The solicitor general’s office, though it has exclusive access to internal government data, “is not a neutral purveyor of facts,” said Amanda Frost, a law professor at American University. “It can’t be relied upon blindly.”

The solicitor general’s office, which declined to comment for this story, has apologized twice before for mistakes stemming from information it submitted to the court: once for notoriously withholding evidence in the 1944 case Korematsu v. United States, and a second time for submitting uncited evidence in the 2009 ruling Nken v. Holder.

What distinguishes the Demore apology is that the solicitor general’s office directly proposed that the court “amend its opinion to delete” the mistake. The Demore ruling, however, was already made official in Volume 538 of United States Reports.

The court’s “commitment to getting things absolutely right is commendable,” said Richard Lazarus, a Harvard Law School professor. But the practice of quietly tinkering with opinions after the fact, and then being nontransparent about what changes were made, is “fairly indefensible,” said Lazarus, the first legal scholar to document the process.

A four-step revision process

The four-step editing process begins with the bench opinion, the version that is announced day-of from the bench and widely read by journalists and legal scholars. Days later the slip opinion appears, correcting mistakes brought to the court’s attention during the post-decision scrutiny. Years later come the preliminary prints, essentially dry runs for United States Reports.

Then, at last, the bound volumes of United States Reports themselves are published. It usually takes a set of three to five volumes to collect just one term’s worth of court opinions, orders, and case tables.

All the while, as the court strives for editorial accuracy from its private chambers, the old, unrevised opinions still float around the Internet unchanged.

“It’s a quality control problem,” said Jack Metzler, a Washington appellate attorney who found the court’s internal style guide and published it in book form earlier this year. “The more sources that are out there for Supreme Court opinions, the harder it is to make sure all those documents have the right information.”

Since Lazarus began writing about the editing practice, tech-savvy lawyers have developed workarounds to discover the justices’ edits, in lieu of stronger transparency from the court., a free online legal database, uses cryptographic hash to detect changes to opinion PDFs on the court’s website. If a PDF changes in any way, CourtListener downloads the altered file but also backs up the original, so the public can judge the difference for themselves.

“Most of the revisions are super mundane, probably 99 percent,” said Mike Lissner, a co-founder at the Free Law Project, which created CourtListener.

Of course, no landmark decision will crumble because of a misplaced apostrophe.

“What did strike me as a big deal is the Supreme Court never told anyone they were making these changes,” said David Zvenyach, who created a software program called SCOTUS Servo that tracked changes to opinion PDFs.

Zvenyach wrote a scraper script that visited the court’s website every five minutes to check for edits. If his program made any discoveries, he would tweet them out. Just a couple hundred lines of code “resolved a pretty big structural problem,” Zvenyach said.

Court adapts, slowly

The court, in its way, has slowly adapted. Justice Ruth Bader Ginsburg was the first to formally cop to a blunder: the court’s Public Information Officer Kathleen Arberg informed a select group of reporters via email about an error in Ginsburg’s dissent to Veasey v. Perry in 2014.

The email showed the exact language that was being deleted and replaced, also noting “small stylistic changes on pages 2 and 4.” Then, at the beginning of the last term, the court announced it would annotate slip opinions to reflect any edits.

“It’s okay to admit your mistakes. You enhance your credibility by admitting mistakes,” Lazarus said. “I think the court has figured that out.”

But the court continues to keep private all of its edits past the slip opinion phase, meaning that “any changes that are happening in the bound volumes won’t be discovered until potentially years later,” said Zvenyach.

Peter W. Martin, a former dean of the Cornell Law School, attributes the long wait to “bad habits, reinforced by staffing and funding limitations.” Indeed, the bound volumes are published at a slower rate today than they were in the 1800s, when the court had a heavier caseload and worked without computers.

In addition to the official four-step process, the court reserves into perpetuity one final laver of authority. Right after the title page of a new bound volume, the court can print a list of corrections to previous volumes. It’s called an errata sheet, and the court has no known restrictions on its use. An erratum once corrected the misspelling of “Braislford” to “Brailsford” 204 years later.

Opinions, thought to be solid as the marble steps leading up to the Court, are in perpetual flux.

Bryan A. Garner, editor-in-chief of Black’s Law Dictionary, wonders if there ought to be a statute of limitations on errata sheets: “Going back 13 years is, you would think, stretching the limit.”

Errata sheets are for typos, Martin said, “not something that’s part of the essential underpinning of an opinion,” as in Demore.

The court can always revisit the precedential weight it applies to Demore in the frame of a new case. With Jennings v. Rodriguez, the court will get that chance. Jennings focuses on part of the issue addressed in Demore, specifically the constitutionality of holding illegal immigrants without bail.

Demore found that mandatory detention was constitutional, partly because the solicitor general’s brief argued, incorrectly, that average detention times were short.

Now, the solicitor general, relying on the precedent in Demore, will fight to uphold mandatory detention past the six-month mark in Jennings. The Obama administration wants to maintain its ability to detain illegal immigrants in proceedings for deportation, arguing they pose a flight risk and potential threat to public safety.

But the solicitor general’s brief again includes untested government statistics. The Demore apology, submitted the same day, reads, “the government recognizes its special obligation to provide this court with reliable and accurate information at all times.”

Correction: An earlier version of this piece misstated the middle initial of Peter W. Martin, former dean of the Cornell Law School.