In 1996, riding a wave of public opinion after the Oklahoma City bombing, Congress passed the Antiterrorism and Effective Death Penalty Act1 (AEDPA).2 Among other things, AEDPA heightened the standard a federal habeas court must use when reviewing a state prisoner’s claim already “adjudicated on the merits in State court.”3 Instead of de novo review, the amended 28 U.S.C. § 2254(d) requires these claims to be dismissed unless the state adjudication “resulted in a decision that” either is “contrary to, or involved an unreasonable application of, clearly established Federal law” or is “based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.”4 The Supreme Court has interpreted these provisions to require significant deference to even incorrect state court rulings.5
Prior to AEDPA, the Court in Ylst v. Nunnemaker6 considered a case where state habeas courts had affirmed without opinion a state court’s direct-review denial of a petitioner’s claim on grounds of a state procedural default.7 The Ylst Court held that federal courts should “look through the subsequent unexplained denials [of habeas] to [the direct-review] opinion”8 and employ a presumption, rebuttable by “strong evidence” to the contrary,9 “that a later decision rejecting the claim did not silently disregard that [procedural] bar and consider the merits.”10 But after AEDPA, the Court in Harrington v. Richter11 considered a case where there was only a one-sentence summary order accompanying a state high court decision and no opinion below.12 The Richter Court held that the unexplained state high court decision rejecting a petitioner’s claim was a decision on the merits for § 2254(d) purposes13 and that a federal court must deny a petitioner’s appeal of that summary decision unless it had “determine[d] what arguments or theories supported or . . . could have supported the state court’s decision”14 and found that “there was no reasonable basis for the state court to deny relief.”15 Left unanswered was whether a Ylst-like presumption applied in § 2254(d) cases where there was a reasoned state lower court opinion below a summary denial.
Last Term, in Wilson v. Sellers,16 the Supreme Court answered that question in the affirmative. Settling a circuit split,17 the Court sided with habeas petitioner Wilson, holding that federal courts reviewing state prisoners’ habeas claims under § 2254(d) should “look through” to the last reasoned state court decision and apply a rebuttable presumption that the unexplained affirmance “adopted the same reasoning.”18 Justice Gorsuch asserted in dissent that Wilson adopted a watered-down presumption that effectively allows courts to apply Richter’s harsh practice of imagining any possible basis for denying relief. But the majority opinion in Wilson both explicitly and implicitly indicated that federal courts should continue to apply a strong presumption that the unexplained decision adopted the lower court’s reasoning. In doing so, Wilson likely restricted Richter’s practice of hypothesizing bases to Richter’s specific procedural posture — that is, to cases where there is no reasoned opinion by any state habeas court — thus limiting the heavy and unnecessary burden this practice places on habeas petitioners.
Marion Wilson was convicted of murder and sentenced to death by a Georgia jury in 1997.19 After exhausting his direct appeals,20 Wilson filed a petition in a Georgia state habeas court, claiming, inter alia, ineffective assistance of trial counsel under the federal constitutional standard.21 The state habeas court denied Wilson’s ineffective assistance claim on the merits after a hearing, explaining its reasoning in an opinion.22 The Supreme Court of Georgia denied Wilson’s appeal without opinion, and the United States Supreme Court denied certiorari.23
Wilson next filed a habeas petition in the U.S. District Court for the Middle District of Georgia, bringing the ineffective assistance claim, among others.24 In denying Wilson’s § 2254(d) claims, the district court judge evaluated the lower state habeas court’s opinion rather than the state appellate court’s unexplained decision.25 After a panel of the Eleventh Circuit looked instead at the state appellate court’s decision,26 the Eleventh Circuit granted Wilson’s petition for rehearing en banc to decide the look-through question.27 Writing for the en banc court, Judge William Pryor agreed with the panel that such cases should be assessed under Richter, finding that neither AEDPA nor Richter required federal courts reviewing unexplained state appellate court opinions to “‘look through’ a summary decision on the merits to review the reasoning of the lower state court.”28 inapposite, explaining that Ylst made more sense where the federal court was assuming only that the state higher court had adopted “the same general ground — that is, a procedural ground or on the merits” — than where, as in Wilson, the federal court would have to assume that the state higher court adopted “the same specific reasons provided by the lower court.” Id. at 1236. In separate dissents, Judge Jordan argued that the majority had incorrectly chosen Richter over Ylst when neither squarely controlled the issue,29 and Judge Jill Pryor argued that Supreme Court precedent and federalism favored a look-through presumption.30
The Supreme Court reversed and remanded.31 Justice Breyer wrote for the majority,32 “hold[ing] that the federal court should ‘look through’ the unexplained decision to the last [reasoned] state-court decision” and “presume that the unexplained decision adopted the same reasoning.”33 But the presumption is not absolute: to “rebut the presumption,” the State must “show that the unexplained affirmance relied or most likely did rely on different grounds than the lower state court’s decision, such as alternative grounds for affirmance that were briefed or argued to the state supreme court or obvious in the record it reviewed.”34 This conclusion, the Court wrote, “ha[d] parallels in [the] Court’s precedent,” namely Ylst, which had approved a similar look-through presumption for federal courts trying to determine whether a state summary denial rested on state procedural default.35 The Court quoted language from Ylst discussing the “difficult[y] and artificial[ity]” of attributing meaning to unexplained state appellate decisions in state habeas cases.36 Citing the consensus among the circuits before the Eleventh Circuit’s opinion below, the Court endorsed the look-through presumption as “often realistic, for state higher courts often (but certainly not always) write ‘denied’ or ‘affirmed’ or ‘dismissed’ when they have examined the lower court’s reasoning and found nothing significant with which they disagree.”37 And, the Court noted, “a ‘look through’ presumption is often (but not always) more efficiently applied than a contrary approach — an approach, for example, that would require a federal habeas court to imagine what might have been the state [higher] court’s supportive reasoning.”38
In response to the State’s position that Richter controlled and thus required federal courts to evaluate the summary decision itself, affirming when there was any reasonable basis for the ruling,39 the Court found that “Richter does not control here”40 for three reasons: First, “Richter did not directly concern the issue” in Wilson because “there was no lower court opinion to look to” in Richter.41 Second, Richter did not explicitly preclude the application of Ylst when there was a reasoned lower court opinion and in fact cited Ylst as an example of a case where the presumption that a court’s decision was on the merits may be overcome.42 And third, the Court had already “‘looked through’ to lower court decisions in cases involving the merits,” including one decided the same day as Richter.43
The Court rejected four further arguments made by Georgia. First, in response to the State’s argument “that there could be many cases in which a ‘look through’ presumption does not accurately identify the grounds for the higher court’s decision,” the Court reiterated that the look-through presumption was rebuttable, noting that “the unreasonableness of the lower court’s decision itself” could “provide some evidence that makes it less likely the state supreme court adopted the same reasoning.”44 Second, the Court found the State’s and the dissent’s appeal to “traditional rules of appellate practice”45 misplaced; the Court distinguished the habeas context, where looking through does not create binding precedent and serves only the “specific and narrow purpose” of “identify[ing] the grounds for the higher court’s decision, as AEDPA directs [federal habeas courts] to do.”46 Third, the Court rejected the State’s argument that the look-through presumption “shows disrespect for the States,”47 explaining that its approach was “more likely to respect what the state court actually did” and reiterating again that “the federal habeas court is free” to find the presumption rebutted “[w]here there are convincing grounds to believe the silent court had a different basis for its decision than the analysis followed by the previous court.”48 And finally, in response to the State’s concern that the Court’s holding would “lead state courts to believe they must write full opinions” where they would not otherwise have done so, the Court found that the State had offered no evidence of such an effect in the circuits that were already applying a look-through presumption.49 In any case, “given the narrowness of the context,” the Court “[did] not believe” state courts would “feel compelled to do so.”50
Justice Gorsuch dissented.51 AEDPA, he argued, clearly required a federal habeas court to “focus its review on the final state court decision on the merits,”52 and Richter had made clear that an unexplained state appellate court order was such an adjudication.53 Ylst, he believed, was limited to cases where the question was whether or not a ruling rested on an independent and adequate state ground.54 In addition to finding that the Court’s holding “defie[d] traditional rules of appellate practice,”55 which he did not find distinguishable in the habeas context,56 Justice Gorsuch disputed the Court’s contention that its approach was realistic, noting that the Georgia Supreme Court had recently made statements to the contrary.57 Justice Gorsuch found solace in his conclusion that the Court had “reshape[d] radically petitioner’s proposed presumption before adopting it.”58 He interpreted the Court’s statement that “it is more likely that a state supreme court’s single word ‘affirm’ rests upon alternative grounds where the lower state court decision is unreasonable”59 to mean that “the ‘look through’ presumption can be rebutted ‘where the lower state court decision is unreasonable.’”60 This led, Justice Gorsuch argued, to the same result as if the Court had directly followed Richter: “[A] federal habeas court is neither obliged to look through exclusively to the reasons given by a lower state court, nor required to presume that a summary order adopts those reasons.”61
Contrary to Justice Gorsuch’s interpretation, both the Court’s explicit holding and indications implicit in its opinion reveal an intent, not to “drop us back where we began,”62 but rather to maintain a presumption with teeth. In doing so, the Wilson Court limited one of the harshest pieces of Richter’s legacy — the practice of courts imagining all possible bases for denying relief — to Richter’s specific procedural posture, thus sparing habeas petitioners from a burden that AEDPA need never have imposed on them.
To start, “a good rule of thumb for reading [Supreme Court] decisions is that what they say and what they mean are one and the same.”63 The explicit language of the Wilson presumption itself creates a high bar over which states must jump. Wilson held that, to rebut the presumption, a state must “show that the unexplained affirmance relied or most likely did rely on different grounds than the lower state court’s decision.”64 The Court’s examples of such grounds were those “briefed or argued to the state supreme court or obvious in the record it reviewed.”65 This is a stricter requirement than Justice Gorsuch admitted to finding in the majority opinion: rather than requiring federal habeas courts to “sustain a state court summary decision denying relief if [the record below] reveal[s] a basis to do so reasonably consistent with this Court’s holdings,”66 Wilson requires specific and concrete evidence of alternate grounds on which the state appellate court “relied or most likely did rely.”67 Moreover, the Court described the look-through presumption as preferable to a “contrary approach . . . that would require a federal habeas court to imagine what might have been the state court’s supportive reasoning,”68 that is, the Richter approach of hypothesizing all possible bases for relief.69
The Court’s responses to the State and the dissent later in the opinion do not relax this standard. In response to concern that a look-through presumption would not account for situations in which the state high court’s summary decision did not in fact adopt the lower state habeas court’s reasoning, the Court noted again that the presumption was rebuttable and stated that “the unreasonableness of the lower court’s decision itself provides some evidence that makes it less likely the state supreme court adopted the same reasoning.”70 But that evidence alone would be insufficient to rebut the presumption: in the subsequent sentence, the Court stated that “additional evidence” — that is, evidence in addition to the fact of a lower state court’s unreasonableness — “that might not be sufficient to rebut the presumption in a case like Ylst would allow a federal court to conclude that counsel has rebutted the presumption.”71 As an illustration of such additional evidence, the Court used the same example as it did in its holding: “[c]onvincing alternative arguments for affirmance made to the State’s highest court or equivalent evidence presented in [the State’s] briefing to the federal court.”72 So the Court cannot have meant “that essentially only reasonable (and so sustainable) lower state court opinions are presumptively adopted by final state court summary decisions,”73 as Justice Gorsuch claimed. Instead, the Court affirmed that, while the Wilson presumption may be rebutted in situations where the Ylst presumption would not, rebuttal requires at least evidence of the state higher court likely relying on different grounds, and perhaps also evidence of the state lower court’s unreasonableness. Thus, the passage relied on by Justice Gorsuch to show the weakness of the Wilson presumption in fact shows its strength.
Other elements of the majority opinion implicitly indicate the strength of the Wilson presumption. Wilson reversed an Eleventh Circuit opinion that adopted exactly the position Justice Gorsuch claims Wilson adopted: “[T]hat federal courts need not ‘look through’ a summary decision on the merits to review the reasoning of the lower state court.”74 Moreover, the Court discussed approvingly “the many Circuits that have applied Ylst outside the procedural context,”75 seemingly equating the Court’s own presumption with the approaches taken by the circuits that had chosen to look through.76 These circuits had been applying the presumption almost categorically, without mention of the presumption being rebuttable.77 The circuit court decisions released since Wilson lend further support to this reading: The Sixth,78 Seventh,79 and Ninth80 Circuits, their approach approved by Wilson, have continued to employ a straightforward look-through presumption. The Eleventh Circuit, on remand in Wilson, applied the Supreme Court’s holding in a similarly straightforward manner, without mention of the presumption being rebuttable.81 If there was any doubt that Wilson meant to maintain a strong presumption, the circuits have laid it to rest.
Arguably the greatest virtue of the Wilson presumption is that it spares habeas petitioners from having to prove not only that the actual state court opinions denying them relief were unreasonable but also that any possible opinion denying them relief would have been unreasonable.82 This harsh standard, created by Richter,83 has been criticized as an unnecessary heightening of AEDPA’s already strict standard.84 And by repudiating this practice in cases where a state higher court summarily affirms a reasoned state lower court opinion denying a state habeas petitioner relief, Wilson must logically also condemn the practice in the “ordinary” § 2254(d) case where there is a reasoned state higher court opinion. If imagining all possible bases for a state higher court’s opinion is unacceptable when that court has produced only a summary decision, how could it possibly be acceptable when that court has explained the grounds on which it decided? Indeed, the Fifth Circuit recently recognized the possibility of this broader effect of Wilson, though it had no occasion to decide the issue.85
Limiting the practice of hypothesizing bases for relief to the procedural posture in Richter — that is, where there is only a state high court summary denial of relief and no reasoned opinion below86 — is consistent not only with Wilson’s explicit repudiation of the practice in its own procedural posture but also with Wilson’s softened characterization of ordinary § 2254(d) review. The Court described § 2254(d) as requiring a federal habeas court to “simply review the specific reasons given by the state court and defer to those reasons if they are reasonable.”87 Absent was the extreme language from Richter that Justice Gorsuch would have used to describe § 2254(d).88 And in addition to later distinguishing Richter as not controlling in cases where there is a reasoned opinion by a lower state habeas court,89 the Court’s initial mention of Richter in its description of ordinary § 2254(d) review reduced Richter to mere support for the proposition that federal habeas courts should “give appropriate deference to” the applicable state court decision.90 Thus, in both logic and tone, Wilson seems to end the Richter practice of imagining all possible bases for relief in all but one specific procedural posture,91 limiting the harsh effects of an unnecessarily strict interpretation of AEDPA.92