Page:Jones v. Hendrix.pdf/55

Rh to override that value (assuming that it actually intended to completely foreclose new legal innocence claims). Congress did not do so; yet the majority reads its silence to accomplish that same extraordinarily inequitable result.

The practical consequences that inure from the majority’s reading also undercut substantially the negative inference upon which the majority relies. We have consistently warned that courts should “resis[t] an interpretation of [AEDPA] that would ‘produce troublesome results’ [and] ‘create procedural anomalies.’ ” Panetti, 551 U. S., at 946 (quoting Castro v. United States, 540 U. S. 375, 380 (2003)). The majority does not speak to this at all, but its interpretation of §2255 produces bizarre outcomes.

First, there is the quirky procedural anomaly that arises due to the fact that statutory innocence claims are fully authorized in the postconviction review context. This Court’s recognition that a statute covers a narrower scope of criminal conduct than was previously acknowledged falls within the narrow subset of criminal law decisions that are fully retroactive, meaning that a federal prisoner can rely upon that new determination whether his case is still on direct review or not. Schriro v. Summerlin, 542 U. S. 348, 351–352 (2004); Bousley, 523 U. S., at 620–621. But reading §2255(h) to bar a successive petition raising legal innocence would mean that most prisoners who would (remarkably) be eligible for such retroactive relief would turn out to have no mechanism for actually requesting it. A strange practical conundrum, to say the least.

Inferring that §2255(h) bars legal innocence claims when brought in a successive petition also produces stunningly disparate results that bear no relationship to Congress’s