Last week we learned that the federal government provided the Supreme Court with bad data about the length of time migrants remain in detention. When? In 2002 as the Court geared up to consider what has become an enormously important decision, Demore v. Kim, 538 U.S. 510 (2002). In Demore, the Court held that the Immigration and Nationality Act’s so-called mandatory detention provision, § 236(c), is constitutionally permissible. For fourteen years, lower courts have dutifully followed the Court’s instruction.

Now we know that the Court’s reasoning turned on wrong statistics. In its 2002 opinion, the Court cited data from the Executive Office for Immigration Review (EOIR), the Justice Department unit that includes the nation’s immigration courts and the Board of Immigration Appeals. Those data, according to the majority opinion written by the late Chief Justice Rehnquist, indicated that “the detention at stake under §1226(c) [INA § 236(c)] lasts roughly a month and a half in the vast majority of cases in which it is invoked, and about five months in the minority of cases in which the alien chooses to appeal.”

It turns out that back then it actually took 113 days, on average, for an immigration judge to decide a case involving a migrant subject to mandatory detention. Letter from Ian Heath Gershengorn, Acting Solicitor General, to the Honorable Scott S. Harris, Clerk, U.S. Supreme Court 3 (August 26, 2016). When a detained migrant appealed, it took 382 days, not five months, for the case to be completed. Id.

Clearly these are meaningful differences. The critical question is what now? I suggest that the Court revisit Demore entirely. It could look for an opportunity to revisit Demore in a subsequent case. Such a case is already on the Court’s docket, Jennings v. Rodriguez. Arising from the Ninth Circuit, Jennings asks the Court to review the procedures that have developed across the country to review claims of prolonged detention—limits that arose directly from Justice Rehnquist’s recognition in Demore that detention “lasts roughly a month and a half.”

Another possibility exists. The Court could revise the opinion to remove the reasoning that turns on this damning data. Though it’s unusual for the Court to make changes to preexisting opinions, it’s not unprecedented. After painstaking comparisons of multiple versions of Supreme Court opinions, Richard Lazarus found that Justice Taney added approximately eighteen pages to his infamous majority opinion in Dred Scott v. Sanford, 60 U.S. (19 How.) 393 (1857), upholding the principle of separate-but-equal. Richard J. Lazarus, The (Non)Finality of Supreme Court Opinions, 128 Harvard Law Review 540, 544 (2014).

Though most changes happen within months or a few years of the decision being issued, sometimes they happen much later. According to Lazarus, errata sheets have been used to correct all type of errors “from decades earlier, or more.” Id. at 574. In one instance, he found a change made in 2013 to a case published in 1991—twenty-two years later! And though substantive revisions are in the minority, they “do not appear to be unusual.” Id. at 569. Sometimes the Court has added an important word like “not” to a phrase—thus flipping the sentence’s meaning—and at other times it has deleted entire sentences. Id.

Meanwhile, advocates will likely begin to challenge lower court decisions relying on the federal government’s poor statistical representations. If the Supreme Court insists on going on as if it hadn’t received this distressing information about the federal government’s misrepresentation, I expect that advocates will eventually force the issue to the Court one lower court decision after another. Instead of that prolonged path, I hope the Court chooses to resolve this predicament on its own.