Taming Cerberus: The Beast at AEDPA's Gates

Introduction

The Antiterrorism and Effective Death Penalty Act of 19961 (AEDPA) established the current regime under which federal courts address petitions for a writ of habeas corpus by state prisoners. Riddled with ambiguities, AEDPA has frustrated judges and commentators alike.2 Because Congress either failed to conclusively resolve or—more likely—did not even consider the text’s application to a multitude of intricate scenarios, judges fall back on three considerations that animate federal habeas jurisprudence to construct the AEDPA regime: finality, comity, and federalism.3 Currently without limitations, this three-headed beast wreaks havoc, upending traditional methods of statutory interpretation and neutral decision-making.4 To resolve difficult questions of habeas jurisprudence, federal judges must first tame Cerberus.5

This Comment advocates a more nuanced approach to those three principles within the context of an issue that the Supreme Court will decide this coming Term in Wilson v Sellers.6 The problem at hand arises when a higher court—usually the state supreme court—issues a summary disposition affirming the decision of a lower court whose written opinion rejected the merits of the claim.

Marion Wilson Jr took a long, circuitous route from his state-court conviction to the Eleventh Circuit en banc in Wilson v Warden, Georgia Diagnostic Prison7 and finally to the Supreme Court in Wilson v Sellers.8 As federal habeas corpus is a form of collateral review, Wilson is challenging a final judgment from a Georgia state court that sentenced him to the death penalty. Wilson needed to exhaust available state remedies9 before accessing a federal forum in which to claim he is being held “in violation of the Constitution or laws or treaties of the United States.”10 Exhaustion required pursuing his claim—that his attorney’s investigation of mitigation evidence at the penalty phase of his trial constituted constitutionally ineffective assistance of counsel11 —through at least one chain of the state appellate system.12 Claims of ineffective assistance of counsel are typically initially presented in state postconviction review (as opposed to on direct review),13 and Wilson’s story is no different. Faced with a “labyrinth”14 of procedural requirements, he nonetheless properly presented his federal claim on state post­conviction review to the Superior Court of Butts County, which issued a written opinion denying relief on the merits.15

In Georgia’s postconviction review system, the petitioner must seek a certificate of probable cause (CPC) to appeal a denial of habeas relief.16 There are two dimensions to the resulting decisions: the legal significance vis-à-vis the petitioner’s claim and the content justifying the result. As to the first, the Georgia Supreme Court’s decision to deny the CPC is a merits determination.17 Thus, in Georgia, a decision not to hear the claim affirms the decision below, as opposed to, say, the Supreme Court’s decision to deny a petition of certiorari, which leaves the lower court’s decision in place without affirming the merits.18 In technical terms, Georgia’s system is one of nondiscretionary review, while the Supreme Court and most other state courts of last resort provide discretionary review.19 And, as relevant to the second dimension, the Georgia Supreme Court denied Wilson’s request for a CPC with a summary disposition.20 Summary dispositions are unexplained decisions that typically contain no information beyond the result—in other words, just “granted” or “denied.”

Having exhausted state postconviction review, Wilson filed his claim for habeas relief in federal court.21 At issue was the application of AEDPA’s state-federal relitigation bar, 28 USC § 2254(d). Section 2254(d) directs federal habeas courts to review “the last state-court adjudication on the merits.”22 This relitigation bar precludes review of the merits of the underlying claim unless the petitioner demonstrates that the state-court decision is unreasonable.23 The state’s chosen structure for postconviction review determines whether the written opinion or the summary disposition was the last adjudication on the merits. If review was discretionary—and thus not on the merits—there is controlling Supreme Court precedent: the court should “look through”24 the summary disposition to evaluate the last written opinion.25 Because the written opinion is the last adjudication on the merits, it is the operative decision for § 2254(d).

But, in states like Georgia that provide nondiscretionary review, the denial of review—here, the decision not to issue a CPC—is an adjudication on the merits. Thus, the summary disposition is the operative decision for § 2254(d).26 If Georgia summarily denies an appeal from a lower court that issued no written opinion, Supreme Court precedent dictates that the federal court review the record for potential reasons that could have supported the result of the summary disposition,27 which this Comment calls “hypothesizing.” However, when Georgia summarily denies an appeal from a lower court that did issue a written opinion, the form of review by the federal court is contested. The parties agree that § 2254(d) operates over the summary disposition. They contest, however, whether the silent denial presumptively adopts the rationale of the lower-court written opinion, such that judges should not supply hypothetical reasons.

Two lines of Supreme Court precedent present equally plausible but incompatible responses (looking through and hypothesizing) to this sequence of state decisions (summary disposition affirming a written opinion). The en banc Eleventh Circuit, by a slim 6–5 margin, held that federal courts should hypothesize reasons that could have supported the summary disposition.28 In the process, the court created a circuit split with the Fourth and Ninth Circuits, who opted to look through to the reasoning in the lower-court opinion.29 As mentioned above, the Supreme Court granted Wilson’s petition for certiorari to settle the look-through/hypothesizing debate.30

The issue is well deserving of its spot on the Court’s docket. There are real stakes in the choice between looking through and hypothesizing for the petitioner, with the approach outcome-determinative given two conditions. First, the lower-court written opinion must be unreasonable under § 2254(d). As the concern is locating unreasonably decided claims, the written opinion must fail § 2254(d) for habeas relief to be possible under either approach.31 If the court looks through to the unreasonable lower-court opinion, it will find § 2254(d) satisfied, which, in most cases, dictates that relief will be granted.32 Second, there must exist a reasonable basis for the opinion that satisfies § 2254(d), such that the court will deny relief if it chooses to hypothesize. The petitioner carries a heavy burden when courts hypothesize: summary dispositions are reasonable “so long as ‘fairminded jurists could disagree’ on the correctness of the state court’s decision.”33 Hypothesizing is more deferential than review based on the opinion’s actual reasoning,34 and summary dispositions are a common state-court method of deciding habeas petitions.35 Thus, the subset of cases brought by state prisoners that involve both an unreasonable lower-court opinion and a hypothetically reasonable summary disposition is significant.36 To make the choice stark: In situations when the federal court would have granted the petition for habeas relief due to the unreasonable written opinion, what impact should a later silent denial have?37

Part I reviews the development of merits review after AEDPA, giving special attention to the Supreme Court’s approach to summary dispositions. Part II recounts the development of the circuit split created by Wilson. Then, after Part III  finds that consideration solely of the text of the statute and the case law leaves residual indeterminacy, Part IV faces the triplet heads of finality, comity, and federalism. On close examination, courts’ application of these three principles often prove unintuitive and contradictory. After recasting these abstract values in more concrete terms, the look-through presumption is revealed to be not only legally consistent but also normatively attractive. Armed with a more nuanced approach, the Court can reclaim finality, comity, and federalism from their current status as across-the-board presumptions against state prisoners.

I. Reaching the Merits of State Decisions under AEDPA

Habeas corpus is a “writ employed to bring a person before a court, most frequently to ensure that the person’s imprisonment or detention is not illegal.”38 The Great Writ’s common-law origins stretch back to 1305,39 and the Suspension Clause of the Constitution guarantees its existence,40 at least for federal prisoners.41 Though Congress authorized habeas relief for prisoners in federal custody in 1789,42 state prisoners generally could not bring habeas petitions in federal court until 1867.43 At first, federal courts could grant habeas petitions only if the tribunal was not competent—that is, it lacked jurisdiction to hear the case.44 As the Warren Court incorporated the protections of the Bill of Rights against the states, federal habeas review became an attractive means of ensuring the compliance of state criminal justice systems.45 This expansion reached its apex in Brown v Allen,46 which officially rejected the jurisdictional test and directed federal courts to review the merits of constitutional challenges to state convictions.47

The availability of habeas relief contracted in the following decades under the Burger and Rehnquist Courts,48 coming to a head with the Oklahoma City bombings, which catalyzed habeas reform efforts by Congress.49 The resulting 1996 bill, AEDPA, strengthened existing procedural barriers to merits review,50 while adding a novel and particularly unwieldy one. That new restriction was § 2254(d), which contains the state-federal relitigation bar:

An application for a writ of habeas corpus on behalf of a person in custody pursuant to the judgment of a State court shall not be granted with respect to any claim that was adjudicated on the merits in State court proceedings unless the adjudication of the claim—(1) resulted in a decision that was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States; or (2) resulted in a decision that was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.51

Twenty years later, federal courts are still grappling with the fallout from this provision. This Part contains a brief overview of the development of access to merits review after AEDPA. Part I.A discusses the gatekeeping role of § 2254(d), while Part I.B outlines the Supreme Court’s approach to summary dispositions in particular.

A. AEDPA’s Gates

To understand AEDPA’s current gatekeeping functions, one first must picture the prior landscape for obtaining habeas relief. Before AEDPA, a federal habeas court would deny the writ based on state procedural bars and would defer to the state court’s findings of fact, but the court would review legal conclusions de novo.52 In 1992, when Justice Clarence Thomas’s plurality opinion in Wright v West53 suggested that the Court’s precedents implied a more deferential standard,54 Justice Sandra Day O’Connor forcefully disagreed “that a state court’s incorrect legal determination” could be allowed to “stand because it was reasonable.”55 Rather, the Court “ha[d] always held that federal courts, even on habeas, have an independent obligation to say what the law is.”56 After the passage of AEDPA, contemporary commentators interpreted § 2254(d)(1) as “more convincing[ly]” consistent with O’Connor’s position in West, thus minimally impacting merits review.57 Advocates of this view interpreted § 2254(d) as requiring courts to start with the state decision, as opposed to jumping straight into analyzing the validity of the petitioner’s current detention.58 Federal habeas courts would act as de facto appellate courts, focused on reviewing another court’s decision but still deciding questions of law de novo.59

In Terry Williams v Taylor,60 O’Connor, writing for the Court on this issue, repudiated this potential interpretation of § 2254(d).61 Not only must a legal determination be incorrect, but it must also be unreasonable62 —precisely the requirement she had rejected in West. She considered § 2254(d) to be Congress’s attempt to depart from existing law in a direction more consistent with Thomas’s understanding.63 Section 2254(d)(1) outlines a new requirement, furnishing two prongs by which a habeas petitioner can show a decision to be legally unreasonable. The “contrary to” prong is satisfied if the state court “applies a rule that contradicts the governing law” set forth in Supreme Court precedent or “confronts a set of facts that are materially indistinguishable from a decision of [the Supreme] Court and nevertheless arrives at a [different] result.”64 A decision fails the “unreasonable application” prong if “the state court identifies the correct governing legal principle . . . but unreasonably applies that principle to the facts of the prisoner’s case.”65 On both prongs, the question is one of objective reasonableness.66

Terry Williams had enormous consequences for petitioners. AEDPA, for the most part, retained and strengthened preexisting barriers to merits review. For example, the petitioner must first present all of her claims to a state court,67 without falling prey to any state procedural bars.68 The petitioner must comply with the rules for successive petitions69 and the statute of limitations.70 But Terry Williams added an additional “gate” that blocks federal courts from reviewing the merits of the claim. Section 2254(d) has grown only more restrictive over time.71

Though usually described as a standard of review,72 § 2254(d) is better conceptualized as a rule of preclusion.73 The reasons are two-fold. First, after Cullen v Pinholster,74 § 2254(d) review is limited to the record before the state court when the claim was adjudicated.75 Furthermore, after Greene v Fisher,76 the relevant body of law is judged from the moment the last state court reached the merits of the claim.77 Even when subsequently uncovered evidence or a change in decisional law reveals that the petitioner would clearly prevail on her underlying claim, the federal court may consider solely whether the state court “badly fumble[d] merits processing of the legal and factual data before it.”78 Second, while satisfying § 2254(d) is a necessary precondition to relief, it is not sufficient to end merits review. Once the petitioner passes the relitigation bar, she still must prevail on de novo review.79 In sum, § 2254(d) is a “statutory preclusion rule based on the state decision” that measures the fault of the decision-maker, “not a standard for reviewing the merits of the underlying claim.”80

B. Making Sense of Silence

Summary dispositions complicate the application of § 2254(d). AEDPA deference, as expounded by Terry Williams, is “premised on an ideal of reasoned dialogue between state courts and federal courts sitting in habeas.”81 Summary dispositions are particularly ill-suited for this type of test. When a federal court reviews a summary disposition, due to its opacity, the court has difficulty determining whether the result is the product of reasoned deliberation, a “haphazard glance,” or “no deliberation at all.”82 Summary dispositions, however, are a common means of disposing of state postconviction review claims.83 In California, for example, over 97 percent of state habeas claims were decided by summary disposition from 2006 through 2009.84 This Section focuses on the two Supreme Court cases that establish methods by which federal courts attribute reasons to silent decisions.

1. Ylst’s look-through presumption.

Ylst v Nunnemaker,85 decided in 1991, is the origin of the look-through presumption. After his conviction for murder, Owen Nunnemaker argued that Miranda v Arizona86 barred certain psychiatric testimony introduced by the state.87 The California Court of Appeal affirmed the conviction in a written opinion, relying on a state contemporaneous objection rule.88 Nunnemaker’s state habeas petition was denied without opinion at each level of postconviction review, culminating in a summary disposition by the California Supreme Court affirming through a denial of nondiscretionary review.89 When a petitioner’s claim is procedurally defaulted, a federal habeas court cannot reach the merits unless one of the exceptions is satisfied.90 But a state court that reaches the merits of the claim after the default vitiates the procedural bar in federal court.91 In this case, the Ninth Circuit lifted the state procedural bar (the contemporaneous objection rule) “because the California Supreme Court did not ‘clearly and expressly state its reliance on Nunnemaker’s procedural default.’”92 The Ninth Circuit reasoned that it could not be sure that the state supreme court did not actually adjudicate the merits of the federal claim when silently denying relief.93

The Supreme Court reversed, reinstating the procedural bar. The Court first observed the difficulty of ascribing reasoning to a summary disposition. Because members of the court need not even agree on the rationale, sometimes “the basis of the decision is not merely undiscoverable but nonexistent.”94 To ensure “sound results,” the Court adopted an administrable and accurate presumption that “[w]here there has been one reasoned state judgment rejecting a federal claim, later unexplained orders upholding that judgment or rejecting the same claim rest upon the same ground.”95 The Court explained why silence should be interpreted as agreement with the reasoning of the lower court:

The maxim is that silence implies consent, not the opposite—and courts generally behave accordingly, affirming without further discussion when they agree, not when they disagree, with the reasons given below. The essence of unexplained orders is that they say nothing. We think that a presumption which gives them no effect—which simply “looks through” them to the last reasoned decision—most nearly reflects the role they are ordinarily intended to play.96

This look-through presumption is not ironclad; rather, it is rebuttable by “strong evidence.”97 In the context of procedural default, federal courts look through to the written opinion unless the petitioner demonstrates that a subsequent court reached the merits of the claim.98 Gesturing at the existence of other exceptions, the Court specifically mentioned “a retroactive change in law [ ] eliminat[ing] that ground as a basis of decision” as strong evidence that a silent decision reached the merits of a claim.99 Though decided five years prior to the passage of AEDPA, the Court recently affirmed Ylst as good law.100 And, in Brumfield v Cain,101 the Court invoked Ylst outside the context of procedural default to look through a denial of discretionary review.102 Neither Ylst (a case applying procedural default) nor Brumfield (a case with discretionary review) involved summary dispositions that reached the merits of the petitioner’s claim, and thus they do not resolve Wilson on their own terms.103 The next Section introduces the advent of hypothesizing as the main alternative to the look-through presumption.

2. Richter deference: Hypothesizing.

In Harrington v Richter,104 the Supreme Court examined the direct intersection of § 2254(d) with summary dispositions. Joshua Richter was sentenced to life without parole for, among other things, murder, and the conviction, affirmed on appeal, became final after the California Supreme Court denied a petition for review.105 He then filed his original petition for state postconviction review in the California Supreme Court, asserting ineffective assistance of counsel, which was rejected in a one-sentence summary disposition.106 Richter’s next stop was federal district court, whose denial of relief was upheld by a three-judge panel of the Ninth Circuit.107 The Ninth Circuit then granted rehearing en banc, reversing the district court’s decision.108 In so doing, the en banc court “questioned whether 28 USC § 2254(d) was applicable to . . . a summary denial . . . but it determined the [ ] decision was unreasonable in any event.”109

Before Richter, circuit courts were divided on how to defer to summary dispositions,110 and many commentators had argued that summary dispositions should not receive AEDPA deference, either because they are not “adjudications on the merits”111 or because they are per se unreasonable applications of federal law.112 The Supreme Court first resolved that § 2254(d) applies to summary dispositions, as “[t]here is no text in the statute requiring a statement of reasons.”113 Section 2254(d) “refers only to a ‘decision,’ which resulted from an ‘adjudication.’”114 Therefore, a state court need not write an opinion to take advantage of AEDPA deference.115 The Court next explained how § 2254(d) applies to summary dispositions. It cited the Terry Williams standard116 but delivered a novel formulation of AEDPA deference:

Under § 2254(d), a habeas court must determine what arguments or theories supported or, as here, could have supported, the state court’s decision; and then it must ask whether it is possible fairminded jurists could disagree that those arguments or theories are inconsistent with the holding in a prior decision of this Court.117

In effect, federal courts start with the result (denial of relief) and hypothesize potential reasons that would satisfy § 2254(d). Not only must the petitioner show that the result is incorrect, but also that its incorrectness is undebatable, such that all fairminded jurists would agree. To reach this result, the Court invoked Cerberus, claiming that finality, comity, and federalism support hypothesizing.118

The Richter Court was unperturbed by “the theoretical possibility that the members of the California Supreme Court may not have agreed on the reasons for denying his petition,” dismissing it as “pure speculation.”119 The Court cited Ylst for the proposition that “the mere possibility of a lack of agreement” does not “prevent[ ] any attribution of reasons to the state court’s decision.”120 Acknowledging the heavy burden on the petitioner to refute all hypothetical reasons supporting the summary disposition, the Court remarked that “[i]f this standard is difficult to meet, that is because it was meant to be.”121 Finding that the summary disposition “required more deference than it received,” the Court reversed the Ninth Circuit’s grant of habeas relief.122

Richter is animated by the concern of burdening state courts with opinion-writing requirements.123 The ability to not issue a written opinion “preserve[s] the integrity of the case-law tradition” and can “enable a state judiciary to concentrate its resources on the cases where opinions are most needed.”124 The Court dismissed the possibility that such broad deference “will encourage state courts to withhold explanations for their decisions.”125 Along these lines, the Court later issued a prohibition in Johnson v Tara Williams126 against “mandatory opinion-writing standards on state courts.”127

II. At the Intersection of Ylst and Richter

The preceding Part introduced two techniques for reviewing summary dispositions: the look-through presumption and hypothesizing. This Part examines how courts have approached the choice between these two techniques for a particular subset of cases. This Comment is concerned with the domain of cases involving silent denials of nondiscretionary review in which there is a lower-court written opinion adjudicating the merits of the claim. Table 1 presents a typology of decisions in state court that determine the application of § 2254(d).

Category A is the paradigmatic AEDPA case—the type to which § 2254(d) is tailored.128 Terry Williams represents the operative test, by which federal courts defer to the reasoning in the state-court decision rather than just the result. While some argue deference to the result is mandated in a post-Richter world,129 the prevailing consensus is that Richter did not abrogate Terry Williams sub silentio.130

Category B contains denials of discretionary review, which operate much like denials of certiorari by the Supreme Court. A denial of discretionary review is not an “adjudication on the merits,” and so falls outside the reach of § 2254(d).131 In Brumfield, the Supreme Court reviewed a denial of discretionary review that left in place a written opinion by the lower state court.132 Because the Court could not defer to the silent denial, it looked through, applying § 2254(d) directly to the lower-court opinion.133 As succinctly put by the Eleventh Circuit–appointed amicus in Wilson, the Brumfield Court was not quite “‘looking through,’ but rather ‘looking for’ the one operative adjudication on the merits.”134

Category C, on the other hand, is Richter redux. When there is no written opinion at any level of review, looking through is impossible. Denial of nondiscretionary review is a merits determination, so, faced with the choice between no deference and maximal deference, the Court chose the latter, creating a regime in which § 2254(d) sometimes requires deference to the result.135

Category D consists of Ylst-type cases. When a federal habeas court reviews a denial of nondiscretionary review, it has to look to the lower-court opinion to determine whether the claim was decided on the merits or procedural grounds. If the written opinion rested on a state procedural rule, then the federal court applies the judge-made doctrine of procedural default.

Category E, which combines elements of B and C, is the focus of this Comment. Federal habeas courts find themselves in Category E when a denial of nondiscretionary review affirms a written opinion that adjudicated the merits of the petitioner’s claim. While both B and C leave open only one avenue for making sense of silence, a court faced with E can choose between the look-through presumption and hypothesizing. In the aftermath of Richter, some commentators cursorily assumed that federal habeas courts would look through summary dispositions to a lower-court opinion whenever possible.136 After all, courts already must review the written opinion to check for procedural bars.137 But, as the Supreme Court has not yet squarely faced this question, the circuits were left to exercise their own best judgment. Part II.A catalogues those circuits opting to look through, while Part II.B covers the Eleventh Circuit’s choice to extend Richter deference in Wilson.

A. Circuits That Look Through

The Ninth Circuit, in Cannedy v Adams138 (“Cannedy I”), was the first circuit court to explore the interplay of Ylst and Richter. After being convicted of child molestation, Earl Cannedy hired a new lawyer and moved for a new trial asserting ineffective assistance of counsel.139 The motion for a new trial was denied, so Cannedy simultaneously appealed his conviction to and filed a petition for a writ of habeas corpus from the California Court of Appeal, which issued an unpublished opinion affirming the convictions and denying the writ.140 Cannedy then filed a petition for review and a “mostly duplicative” petition for a writ of habeas corpus in the California Supreme Court, which declined to review his convictions and summarily denied the writ.141

Cannedy petitioned the Central District of California for a writ of habeas corpus. The district court held, first, that counsel’s performance was constitutionally deficient because the evidence corroborating the victim’s motive for fabrication “was so significant and potentially exculpatory that any reasonable attorney would have sought to admit [it]” and, second, that the deficient performance prejudiced Cannedy.142 Accordingly, the district court granted the writ, and the state appealed.143 The Ninth Circuit began by examining the summary disposition, finding that review was nondiscretionary.144 Therefore, the court could either look through or hypothesize potential reasons.

The Ninth Circuit opted for the Ylst presumption, looking through the summary disposition to the written opinion by the California Court of Appeal.145 The panel majority rejected the dissent’s desire to hypothesize as relying on “an overly broad reading of Richter.”146 Because looking through to the lower court’s written opinion to apply § 2254(d) was a common practice among federal courts before Richter, the majority thought it “unlikely that the Supreme Court intended to disrupt this practice without making its intention clear.”147 Instead, Richter is limited to the scenario in which there is “no reasoned decision at all.”148 Judge Andrew J. Kleinfeld, writing in dissent, disagreed, arguing that Richter—a “sharp rebuke to [the Ninth Circuit’s] previous practice”—controls the interpretation of all summary denials on the merits.149 By refusing to hypothesize reasons supporting the summary disposition, the majority “omit[s] the phrase ‘or, as here, could have supported’” from Richter.150 Thus, Kleinfeld argued that summary dispositions can be supported by hypothetical reasons, even when silently affirming a lower-court written opinion.151

Cannedy I involved a complicating factor: the evidentiary record. The majority reviewed the written opinion rendered by the California Court of Appeal against the “materially improved” record before the California Supreme Court.152 This choice drew criticism from the panel dissent.153 In his dissent from denial of rehearing en banc, Judge Diarmuid O’Scannlain likewise castigated the majority for “declar[ing] a state court’s analysis unreasonable based on evidence not before it.”154 The panel majority had given short shrift to the argument, only pausing to assert that “[h]ad the state supreme court intended different reasoning because of the newly added facts, the court could have provided it.”155

O’Scannlain’s approach differed from the panel majority’s and Kleinfeld’s, as he was willing to accept a background rule that federal courts generally should look through summary dispositions. As discussed earlier, the look-through presumption is rebuttable. The Ylst Court identified an intervening change in law as one such exception.156 O’Scannlain noted, “It requires only a short leap to conclude that . . . where a subsequent change in the facts eliminated the bases for the court of appeal’s decision, the look-through presumption should have been disregarded.”157 This argument is consistent with an alternative account of Cannedy I: the majority was right to initially apply Ylst to the summary disposition, but it erred in failing to find the presumption rebutted by the significant change to the record.158 Once the look-through presumption is rebutted, the court would apply Richter deference.159

In 2016, Grueninger v Director, Virginia Department of Corrections160 aligned the Fourth Circuit with the Ninth. Eric Grueninger, arrested for sexual abuse, gave a confession without an attorney present, days after requesting one.161 His attorney later failed to file a timely motion to suppress the confession, and Grueninger was convicted.162 On state habeas review, Grueninger argued, among other things, ineffective assistance of counsel. The Hanover Circuit (the state habeas trial court) dismissed the petition, rejecting the ineffective-assistance claim on the merits.163 The Supreme Court of Virginia then summarily denied Grueninger’s petition for appeal on the grounds that it contained “no reversible error.”164 This decision was a denial of nondiscretionary review.

Next, Grueninger filed an unsuccessful federal habeas petition, raising similar claims, in the Eastern District of Virginia, after which the Fourth Circuit granted a partial certificate of appealability on the ineffective-assistance claim.165 To review the summary disposition, a unanimous panel of the Fourth Circuit looked through to the Hanover Circuit Court’s opinion, rejecting the state respondent’s argument that Richter demands hypothesizing.166 The court considered Richter bound to its facts: “an original petition” “presented directly to a state supreme court . . . denied by that court in a one-sentence summary order” when “there was no reasoned decision by any state court.”167 “The situation,” it explained, “is different when there is a state-court decision explaining the rejection of a claim.”168 The court “assume[d]” the summary disposition “endorsed the reasoning of the Circuit Court,” and so it applied § 2254(d) to the reasoning in the opinion below.169

The Fourth Circuit, however, found Brumfield to dispel “any doubt about the scope and continued vitality of Ylst after Richter.”170 The court either ignored or glossed over the distinction between nondiscretionary and discretionary review.171 Given the Fourth Circuit’s heavy reliance on Brumfield, Grueninger lends only weak support—perhaps just that of strength in numbers—to the Ninth Circuit’s stance in Cannedy I.

B. A Circuit Split Forms

Wilson is the most recent—and soon to be last—chapter of the look-through saga.172 This issue fractured the en banc Eleventh Circuit, with the majority “conclud[ing] that federal courts need not ‘look through’ a summary decision on the merits to review the reasoning of the lower state court.”173 After being sentenced to death, Marion Wilson Jr filed a state habeas petition asserting ineffective assistance of counsel at the penalty phase.174 The Superior Court denied the petition in a written order, and the Georgia Supreme Court summarily denied his application for a CPC.175 Importantly, the refusal to issue a CPC was a denial of nondiscretionary review, as the court was required to determine that Wilson’s claim lacked “arguable merit.”176 Wilson’s federal petition for habeas relief was denied by the district court, and a panel of the Eleventh Circuit affirmed, applying Richter and eschewing Ylst.177 The Eleventh Circuit then vacated the panel opinion so that the en banc court could decide whether it should instead apply Ylst.178

The majority’s first task was determining the scope of Richter. As opposed to the Ninth and Fourth Circuits,179 the majority did not consider Richter bound to its facts.180 Ylst, on the other hand, is limited to the proposition that summary dispositions “rest[ ] on the same general ground—that is, a procedural ground or on the merits . . . . But it does not follow that a summary affirmance rests on the same specific reasons provided by the lower court.”181 By looking through to the specific reasons, federal courts run afoul of the prohibition in Tara Williams against “mandatory opinion-writing standards.”182 The state court would have to provide a written explanation to avoid “rubber­stamp[ing] [ ] the opinion below.”183

According to the majority, the look-through approach suffers other infirmities. If decisional law shifted in the interim, “the federal court would assume that the state supreme court willfully ignored the intervening change in law”184 (though Ylst explicitly deemed the presumption rebutted by this scenario).185 The majority argued that § 2254(d) was nonetheless violated, as the federal court must consider the lower-court opinion, summary disposition, and briefing before the state supreme court to determine whether the presumption applies. In the majority’s eyes, this constitutes “reviewing the entire process,” not the “decision” or “single adjudication” envisioned by § 2254(d).186 Though unacknowledged by the majority, federal habeas courts have a “duty . . . to determine the scope of the relevant state court judgment.”187 In order to determine what the summary disposition actually decided, the federal court can properly consider the lower court’s written opinion and appellate briefing—part of the state-court record—to interpret the summary disposition.188

III. Residual Indeterminacy

To analyze the competing claims of the look-through presumption and hypothesizing, Part III.A starts with the text of § 2254(d), while Part III.B studies the Supreme Court’s case law. These traditional sources of interpretative material fail to persuasively settle the dispute; neither the text of the statute nor the Court’s precedent dictate a single resolution. Given that the answer is indeterminate, the bedrock first principles of habeas corpus likely will hold sway. Therefore, the next Part develops a more nuanced perspective on the values underlying AEDPA interpretations—finality, comity, and federalism—to conclude that the look-through presumption is both legally consistent with and normatively attractive in light of the Court’s habeas jurisprudence.

A. The Text of § 2254(d)

Supporters of hypothesizing offer a few textual arguments for quarantining Ylst to the context of procedural default. First, they argue that there is “no basis in [§ 2254(d)] . . . for two divergent analytical modes.”189 On this view, there is no textual hook for shrinking the domain of reasons for summary dispositions that follow written opinions compared to ones that do not. There is one problem, however: that ship has sailed. There is likewise no textual basis for “divergent analytical modes” between written opinions, which under Terry Williams are restricted to the domain of reasons expressly stated, and summary dispositions, which under Richter incorporate the domain of all hypothetical reasons. The debate at hand now concerns the respective scopes of Terry Williams and Richter.

Some weight could be placed on the phrase “involved an unreasonable application of [ ] clearly established Federal law.”190 Arguably, when a written opinion “involved” an unreasonable application, it does not matter that hypothetical reasons exist to justify the result. On this view, Congress has created two ways for written opinions to fail § 2254(d)—a results-oriented “contrary to” prong and a reasoning-oriented “involved” prong—while providing only one to pierce summary dispositions (“contrary to”).191 The “contrary to” route, then, is satisfied only when no hypothetical reason justifies the result. Even if this outcome would be perverse,192 the Court would cede if Congress demanded such a result in the plain text of the statute. Though textually defensible, it is certainly not compelled by the statutory language, and Terry Williams did not interpret the “contrary to” prong of § 2254(d) in this manner.193

The core problem with relying on a completely textual approach is that the issue is not one of pure statutory interpretation. Section 2254(d) supplies a rule of preclusion: do not review the merits unless the decision is unreasonable. Section 2254(d) does not purport to prescribe what a state decision means; rather, it says which type of decisions (unreasonable ones) bypass its gate. State law and practice, not § 2254(d), determine the meaning of summary dispositions.194 As § 2254(d) does not speak on the question of a decision’s content, the look-through presumption and hypothesizing are two extratextual attempts to populate the domain of reasons § 2254(d) considers. Given the problems in charting a solely textual course, this antecedent question of the domain of reasons available necessarily must be answered by recourse to extratextual considerations. The following sections demonstrate that the superior lodestar for sorting between these two established modes of analysis is the presence of a written opinion in the state process, not the issuance of a silent denial.195

B. The Path of the Supreme Court Thus Far

While the Supreme Court has yet to squarely confront the intersection of Ylst and Richter, it will resolve this issue in Wilson v Sellers. This Section analyzes the relevant background cases, concluding that precedent does not dictate an outcome in Wilson this coming Term.

One would think the best evidence in support of the look-through presumption is the Court’s own approach under similar circumstances. As noted by a proponent of the look-through presumption, “the Supreme Court has never (ever) applied Richter . . . to a case involving a reasoned lower-court decision.”196 For example, in Premo v Moore,197 decided the same day as Richter, an Oregon postconviction court had denied the petitioner’s claim as “fruitless,”198 which the Oregon Court of Appeals affirmed in a summary disposition.199 Though the summary disposition was the last adjudication on the merits, the Court did not hypothesize; instead, it analyzed the lower court’s written opinion (though without explicitly invoking Ylst).200 The analysis is feasibly consistent with the Court hypothesizing because the Court found that the lower-court opinion was reasonable under § 2254(d).201 That opinion technically is a hypothetical ground on which the summary disposition could have rested.202 Because the opinion cites neither Ylst nor Richter on that point, proponents of both the look-through presumption and the hypothesizing approach claim Moore as support. Nevertheless, Moore’s complete focus on the written opinion suggests the Court limited the domain of § 2254(d) to the reasons in the written opinion.

Furthermore, in Foster v Chatman,203 the Court, on direct review under 28 USC § 1257(a), analyzed the reasons in the lower-court decision because the state supreme court had issued “an unelaborated order . . . provid[ing] no reasoning for its decision.”204 The dissent even raised the same arguments as the Eleventh Circuit, accusing the majority of “parsing the wrong court’s decision” and “impos[ing] an opinion-writing requirement on the States’ highest courts.”205 Though the Court “rarely grants review” under § 1257(a), “choosing instead to wait for federal habeas proceedings,”206 Justice Samuel Alito noted an increasing “trend” to utilize this alternate route to review.207 Even though the Court does not defer to the state-court decision on direct review, it is unclear why the meaning of the state decision would depend on whether the Court reviews for correctness under § 1257(a) or reasonableness under § 2254(d).208

On the other hand, the Court has extended the scope of hypothesizing under § 2254(d) to silent portions of written opinions. The Tara Williams Court held that a state-court opinion that does not address a federal claim is nevertheless presumed to have adjudicated the claim on the merits.209 When a written opinion is silent with respect to a federal claim, federal habeas courts supply hypothetical reasons.210 Notably, however, the Court has exhibited restraint before resorting to hypothesizing. In Lafler v Cooper,211 the Court considered a two-sentence opinion on the merits of a claim to be “not [ ] quite so opaque” as to require hypothetical reasons.212 Moreover, the Brumfield Court described Richter’s scope in rather limited terms. Richter, the Court explained, “requir[es] [the] federal habeas court to defer to hypothetical reasons [the] state court might have given for rejecting [the] federal claim where there is no ‘opinion explaining the reasons relief has been denied.’”213

Two current justices have made explicit the preferences at which the preceding cases can only hint. When the Court denied certiorari in Hittson v Chatman214 on the question presented in Wilson, Justice Ruth Bader Ginsburg, in a concurrence joined by Justice Elena Kagan, responded to the Eleventh Circuit’s abandonment of its “long” practice of looking through denials of nondiscretionary review.215 Ginsburg and Kagan concurred in the denial because they were convinced that the petitioner would not be entitled to relief under either approach,216 but nonetheless wrote to express that “[t]he Eleventh Circuit plainly erred in discarding Ylst” when it “consider[ed] hypothetical theories that could have supported” the summary disposition.217 Ginsburg first identified the overarching principle in § 2254(d) as “direct[ing] a federal habeas court to train its attention on the particular reasons—both legal and factual—why state courts rejected a state prisoner’s federal claims.”218 When the last adjudication on the merits produces an opinion, “[t]his task is straightforward”: “federal habeas court[s] simply evaluate[ ] deferentially the specific reasons” given in that opinion.219 Ylst is the Court’s response to the “more challenging circumstance” when the last state court issues a summary disposition.220 In Richter, because there was no written opinion to look through to, “the Court had no occasion to cast doubt on Ylst.”221 In sum, Richter stands for the proposition “that where the state court’s real reasons can be ascertained, the § 2254(d) analysis can and should be based on the actual ‘arguments or theories [that] supported . . . the state court’s decision.’”222

IV. The Beast Rears Its Heads

The text of § 2254(d) does not resolve the antecedent question of the domain of available reasons that can justify a decision.223 Moreover, neither Ylst nor Richter, by their holdings, resolves the issue, and the Supreme Court’s practice since Richter plausibly supports either approach.224 To make the discussion more tractable, this Part develops a normative framework to evaluate the relative strengths of each position.

The Court interprets AEDPA in light of its legislative purpose to advance the “principles of comity, finality, and federalism.”225 Professor Lee Kovarsky has described this “interpretive mood” as “a sacred cow of modern habeas jurisprudence.”226 While commentators have questioned the ability to attribute a specific purpose to Congress,227 these normative values define habeas jurisprudence.228 The triad not only currently exerts gravitational pull on judicial decisions, but also has done so for more than two centuries—look to any decade of the US Reports to find Cerberus’s paw prints.229

Accepting these values as interpretive guideposts is just the first step to the challenge of constructing AEDPA’s meaning. Courts must study the interaction of the three purposes in the context of the precise issue.230 Kovarsky helpfully describes legislative purpose as a “geometric vector.”231 “While a provision’s text might disclose a particular purpose’s direction,” he notes, “its length—and therefore value—can only be known by reference to the value of the other purposes with which it competes.”232 Federal courts have abstracted AEDPA’s purposes to such a high level of generality that they serve as a presumption in favor of the state respondent on all contested issues of interpretation.233 When doing so, they ignore Justice Antonin Scalia’s admonition that “[n]o legislation pursues its purposes at all costs.”234 In effect, courts disregard the magnitude of the purposes and assume all three invariably point in the same direction without paying attention to the circumstances under which they conflict. Instead, courts would be better served by more closely analyzing these purposes in their concrete applications, looking to the text of AEDPA and the Court’s precedent for direction on how to temper these otherwise-obstinate principles. This Part demonstrates how a nuanced normative framework can foster a productive discussion and identify the superior approach—here, the look-through presumption.

A. Finality

Federal habeas review, a form of collateral review, inextricably raises concerns of finality.235 In a landmark article,236 Professor Paul M. Bator called federal habeas “[o]ne of the areas of acutest controversy” for the finality of convictions.237 If collateral review is unrestrained, society will incur excessive costs from unnecessary and duplicative relitigation while diminishing the reputation and effectiveness of the criminal justice system.238 As a rule of preclusion, § 2254(d) safeguards the interest in finality while making exceptions for certain problematic cases. Because no form of review can completely eliminate epistemic uncertainty, the legality—and, thus, finality—of the sentence cannot depend on whether the constitutional violation actually occurred.239 The criminal justice system could not function—from the perspective of cost, safety, and deterrence—if a prisoner could keep filing rejected claims until he finds a court that agrees with him.240 For a federal court’s judgment to supplant a state court’s, there must be some additional reason for intervention. Bator would have tethered finality to state process,241 and § 2254(d), after Terry Williams, at first operated to correct only “serious errors.”242 Now state-court fault arguably has replaced blatant error as the necessary precondition to federal review of the merits.243 This Section demonstrates that the existence of state fault is the best predictor of finality yielding to a meritorious claim, which militates in favor of the look-through presumption.

The interest in finality explains the proliferation of AEDPA’s gates limiting the availability of a fresh merits determination,244 but finality itself lends no guidance in the choice between the look-through presumption and hypothesizing. The interest in finality matures upon completion of direct review—that is, once the conviction is final. Later events occurring when the petitioner is collaterally attacking his conviction within the postconviction system neither strengthen nor weaken the value of finality.245 In other words, upon the conclusion of direct review, society has some interest in finality. Denote this as M. Postconviction review may develop countervailing reasons to grant the writ, such as evidence that the decision was incorrect to some level N or unreasonable to some level P or that the prisoner is innocent to Q certainty.246 In a fault-based review system,247 the prisoner does not gain relief by diminishing the interest in finality; instead, he shows the interest in correcting state fault surpasses the interest in finality (such that P>M).

1. An example.

To demonstrate, imagine X and Y are state prisoners whose convictions became final after direct review. In both cases, the lower-court opinion is unreasonable for purposes of AEDPA. But X was summarily denied discretionary review while Y was denied nondiscretionary review. A federal court reviewing X’s claim will look through to the unreasonable opinion and find § 2254(d)(1) satisfied,248 while a federal court reviewing Y’s claim must decide between looking through and hypothesizing. The only relevant difference between the cases of X and Y is the procedural threshold for postconviction review by the state’s highest court (discretionary versus nondiscretionary). Therefore, a court reviewing Y’s claim cannot invoke finality to justify departing from the approach for X. To orient within the relevant cases, the interest in finality is no greater in Wilson than it was in Brumfield, so finality cannot justify deviating to hypothesizing from the look-through presumption.

Instead, because § 2254(d) precludes independent review of the merits unless the state-court determination was unreasonable, the question is whether the provision of nondiscretionary review “cures” the fault stemming from the unreasonable lower-court opinion. As Prisoner X did not receive another opportunity to present the merits of the claim, the fault could not have been cured. For Prisoner Y, the denial of nondiscretionary review is a merits determination, so there are colorable defenses for both approaches. Because another court has passed over the merits, one could argue that summary dispositions accompanied by unreasonable lower-court opinions should be treated no differently from bare silent decisions.249 The natural response is that the provision of nondiscretionary review, standing alone, does not cure the fault from an unreasonable lower-court opinion, especially if the opinion “anchors” the review of the state supreme court. Judges on the reviewing court will treat the written opinion as a “salient starting point for their own thinking” about the claim, which dramatically increases the likelihood that the higher court makes the same unreasonable mistake.250

2. False negatives.

A common way to conceptualize errors is as false positives and false negatives.251 In the context of state prisoners seeking a writ of federal habeas corpus, a false positive is when a federal court mistakenly grants relief to a petitioner who did not in fact have his constitutional rights violated. (Basically, the court says “yes” when it should say “no.”) A false negative is when relief is denied to one who did in fact have his constitutional rights violated. (Now the court is saying “no” when it should say “yes.”)

When Congress passed AEDPA, it had both error types in mind. One way to decrease the number of false positives is to prevent federal courts from reaching the merits in the first place.252 If the court cannot decide the underlying claim, it cannot grant relief. This, in turn, will increase the number of false negatives. From the Batorian viewpoint of habeas, this is a feature, not a bug: finality is a meaningful normative commitment only if a court is willing to deny meritorious claims. Federal habeas corpus is interesting, because, unlike in most contexts,253 the goal is not to minimize the number of errors, but to limit the availability of relief irrespective of the production of false negatives.254

AEDPA visibly baked the notion of false negatives into § 2254(d). To obtain relief, not only must the state-court decision be wrong (the false negative), but to even reach the merits, it must also be unreasonably so. All else equal, the gap between “wrong” and “unreasonable” ensures some consistent minimum level of false negatives. Other gates, such as exhaustion, procedural default, and the second-or-successive-petition bar, similarly derail meritorious claims. Nevertheless, Congress intended federal habeas to translate some modicum of false negatives into true positives upon a finding that the state decision was unreasonable. The question, then, is what sorting mechanism best distinguishes cases Congress considered acceptable false negatives from those it wished to convert to true positives.

3. Fault as organizing principle.

Scholars have identified fault as a principle of habeas jurisprudence that determines which prisoners’ convictions remain final.255 Fault is usually discussed both in terms of the state’s possession and the petitioner’s lack thereof, such that “[o]nly by demonstrating his or her own exceptional blamelessness . . . or the exceptional blameworthiness of the state . . . can a petitioner succeed in securing relief from a federal habeas court.”256 Despite the ubiquity of this formulation, the gates precluding merits review demonstrate greater focus on the state alone.

A petitioner can bypass § 2254(d) by demonstrating the last state court was at fault,257 but not by proving her own lack of fault by current evidence—in other words, the Court has refused to recognize a freestanding innocence claim.258 State remedies must be exhausted unless “there is an absence of available State corrective process.”259 For purposes of procedural default, a petitioner can show “cause” most readily when the state is at fault,260 but not when his attorney errs.261 The exceptions prove the rule: ineffective assistance of trial counsel262 and of state postconviction review counsel, under the exception carved out in Martinez v Ryan,263 are more an indictment of the state for failing to provide adequate representation than any statement of the petitioner’s blamelessness.264

Lastly, the Teague v Lane265 bar on retroactive application of new rules produces false negatives despite the lack of fault on behalf of the petitioner.266 Instead, Teague aligns with a state-centric perspective of AEDPA’s gates. A state can choose to expand the availability of new rules not declared retroactive267 (because the petitioner is not faulty) but is forced to give effect to new rules declared retroactive268 (at which point the state becomes the blameworthy party for failing to comply with norms of constitutional obedience). The one-sided nature of fault explains why the state can deviate from Teague in only one direction.

4. Application to denials of nondiscretionary review.

The preceding examples demonstrate that state fault, rather than lack of petitioner fault, better predicts when a petitioner can bypass the habeas gates—in other words, when the decision can cease to be a false negative. The look-through presumption conforms to this pattern: it allows hypothesizing when intervening events external to the state, such as changes to the applicable law269 or factual record,270 make reliance on the lower-court opinion unlikely, but otherwise holds the state to its faulty decision in the court below. One could argue that an unreasonable lower-court opinion likewise is an extreme intervening event that should sever the causal link, on the grounds that it is implausible to ascribe faulty reasoning to a silent decision.271 There is surface appeal to this argument, but it risks tautology: heads I win, tails you lose. The petitioner must first show the opinion is unreasonable, but once she does, it becomes evidence that the later court could not have relied on it. As an empirical matter, this argument is probably wrong. The provision of an unreasonable rationale, viewed under the deferential guise of appellate review, is more likely to engender reliance (by anchoring the court’s approach) than rejection (through its faulty reasoning).272 As § 2254(d) is a “guard against extreme malfunctions in the state criminal justice systems,” the petitioner’s burden should not become heavier once she demonstrates that the opinion is unreasonable.273

The better question is whether § 2254(d) is structured to permit review of this type of false negative. When a lower-court opinion is faulty and a later court silently passes over the merits, the petitioner can show state fault compounded by an absence of externally observable markers of deliberation. This fault is internal to the state. Because “courts generally . . . affirm[ ] without further discussion when they agree . . . with the reasons given below,”274 federal courts should find that the summary disposition “involved” an unreasonable application of federal law under § 2254(d). When federal courts discard this presumption in favor of maximal deference, they distort § 2254(d) to require a greater showing of fault than is required to rebut the interest in finality.275

B. Comity

Though comity and federalism are distinct principles whose constitutive elements potentially conflict, they are frequently invoked as “a unitary interest in deference to state respondents.”276 Comity, in its pure form, describes deference between two coequal sovereigns, typically in the context of international law.277 Federalism complicates the balance of interests by interposing the supremacy of federal law.278 The Supreme Court’s habeas jurisprudence protects two components of sovereign comity (respecting the state’s interest in administering a criminal justice system and in structuring state postconviction review), as well as two of judicial comity (respecting the meaning of state-court decisions, without imposing opinion-writing requirements). Though comity is usually deployed to deny relief to state petitioners, these components can interact in more complex ways than typically acknowledged, as discussed below.

1. The state’s interest in administering a criminal justice system.

Sovereign comity’s first component is implicated by the state’s choice to imprison a person under its jurisdiction. The state’s interest in administering its criminal justice system will often track the general interest in finality, but finality may yield if the state decides to promote another value.

Danforth v Minnesota279 is a good example of the nuance with which the Supreme Court has approached this aspect of comity. In Danforth, the Court held that state courts can choose to expand relief to a “broader range of constitutional violations than are redressable on federal habeas.”280 In the interest of finality, the dissent would have denied the state the ability to broaden relief.281 Because finality, in this context, is primarily a state interest, the majority decided that “considerations of comity militate in favor of allowing state courts to grant habeas relief to a broader class of individuals than is required.”282

The takeaway is that comity and finality can oppose each other, with states given leeway to pursue interests other than finality in postconviction proceedings. States may produce written opinions despite—or maybe even because of283 —the fact that such opinions are more likely to be found unreasonable under § 2254(d). States may be attuned to the fact that criminal justice systems derive their legitimacy from transparent decision-making and accuracy as well as finality.284 Danforth demonstrates that the vindication of constitutional rights is an important counterweight to finality that the state can pursue through appropriate means.

2. The state’s interest in structuring postconviction review.

Danforth helps illustrate the second component of sovereign comity, which is the respect federal courts are to accord the state’s responsibility for administering postconviction review. The states are considered coequal partners in enforcing the Constitution, so federal habeas courts are to treat the state courts as the primary forum for the vindication of state petitioners’ constitutional rights.285 When a state petitioner raises a federal claim, “[c]omity . . . dictates that . . . the state courts should have the first opportunity to review this claim and provide any necessary relief.”286 To that end, Congress codified the exhaustion requirement.287 Section 2254(d) further “demonstrates Congress’ intent to channel prisoners’ claims first to the state courts.”288 The state adjudication on the merits should be the “‘main event’ . . . rather than a ‘tryout on the road’ for what will later be the determinative federal habeas hearing.”289

The “main event” argument not only denies habeas petitioners the opportunity to supplement the record after the state decision290 but also counsels against allowing state respondents to post hoc rationalize decisions with arguments not made before the state court. The state proceeding is not the “main event” if the state respondent can “try out” arguments, later deemed unreasonable, in state court before advancing remedial hypothetical reasons in federal court.291 Richter is the exception, as the Court was forced to choose between de novo review (in which case the federal habeas proceeding unequivocally would be the “main event”) or review of the result (which risks state respondents post hoc rationalizing decisions in federal court).292 Because de novo review represents a larger incursion on the principles of finality and the state’s interest in administering a criminal justice system, reviewing the result was the better option. Too little deference impermissibly treats state courts as lacking the competence to adjudicate federal claims. Deference tantamount to abdication, on the other hand, offends comity by demonstrating a lack of respect for the states’ ability and duty to effectively vindicate constitutional rights.293 The right balance of deference is needed for the state-court proceeding to be the “main event.”

States are granted considerable flexibility in structuring their postconviction systems so that they can most effectively fulfill the responsibility of vindicating constitutional rights. Comparing California’s postconviction system to Georgia’s—an example deployed in Wilson—demonstrates how the look-through presumption is more consistent with this element of comity.294 California allows habeas petitioners to file an original petition with the California Supreme Court, which is free to dispose of the claim by summary disposition.295 Georgia, conversely, requires petitioners to first seek relief in superior court, in which they are “guarantee[d] [ ] at least one reasoned decision addressing their claims,” before they can appeal to the Georgia Supreme Court, which is free to utilize a silent denial.296 The majority, in effect, transplanted a rule produced by the structural circumstances of the former system—that is, Richter hypothesizing—to address a petition arising from the latter.

The Wilson dissent criticizes this approach as “treat[ing] the reasoned opinion of a Georgia superior court as a nullity merely because the Georgia Supreme Court subsequently rendered a summary decision.”297 One could respond in turn that looking through actually treats the summary disposition as though it were a “nullity,” despite deserving a second layer of deference because a higher state court has passed over the merits.298

Yet both of these arguments are slightly off the mark. If a federal court does not look through, the written opinion is still available as a potential hypothetical reason on which the summary disposition may have relied.299 In contrast, if the federal court looks through, the summary disposition presumptively incorporates the reasoning of the written opinion while preserving the ability to rebut the presumption with strong evidence that the decision rested on alternative grounds. The opportunity to rebut is the additional deference given to the later merits determination. While both sides of the debate can advance color­able claims, the look-through presumption is more consistent with how the state has chosen to distribute responsibility among its courts. If a state dedicates resources to producing a written opinion, one can fairly infer that the fruits of its labor are intended as a focal point for further review.

3. Giving effect to a state decision’s intended meaning.

As mentioned previously, federal habeas jurisprudence recognizes two distinct prescriptions from judicial comity. First, when possible, federal courts are to give effect to the intended meaning of the state court’s decision. Second, federal courts are not to impose opinion-writing requirements on state courts. These two factors are discussed in turn.300

AEDPA deference “reflect[s] a ‘presumption that state courts know and follow the law.’”301 Nevertheless, Terry Williams directs federal courts to review the actual reasons provided by the state court, and the Court has elsewhere treated the state court’s reasoning as the subject of § 2254(d) review.302 Even Richter posits the fiction that summary dispositions actually do rest on specific reasoning, which distinguishes hypothesizing from a pure form of deference to the result.303 The import is clear: federal courts do harm to judicial comity if they disregard the state court’s actual reasoning, even if done to sustain the result.304

Take first the situation in which the last adjudication on the merits is a written opinion. Federal courts disrespect judicial comity and undermine the position of state courts as the primary forum for constitutional claims when they disregard the actual reasons used by the state courts. When considering the harms attendant to habeas review, Bator “could imagine nothing more subversive of a judge’s sense of responsibility, of the inner subjective conscientiousness which is so essential a part of the difficult and subtle art of judging well, than an indiscriminate acceptance of the notion that all the shots will always be called by someone else.”305 Nothing offends judicial autonomy more than a federal court tossing away a state court’s given reasons for a prisoner’s continued detention to present its own theory of the claim.306 Congress could have made federal courts the primary forum for resolving claims in the first instance, but it chose state courts as the “main event.”307 Federal courts upturn this balance when they treat a state decision as tabula rasa for speculation.

Furthermore, this aspect of judicial comity dovetails with one of the Court’s rationales for procedural default. If federal courts bypass state procedural rules with regularity, state courts may become “less stringent in their enforcement,” seeing no reason to rely on rules that federal courts themselves do not respect.308 Likewise, state courts will devote less time to parsing the merits of the claim if a federal court will substitute its own reasoning down the line. The Court has recognized before—and should recognize now—the toxic effect of federal-court indifference to the choices made by state courts.

Ylst further supports the conclusion that federal courts should look for the actual reasons whenever possible. Some have mischaracterized Ylst as directing federal courts to ascertain only the general grounds for a denial—as in, procedural or merits-based.309 When the Court looked through in Ylst, it did not ask only whether the claim was procedurally defaulted. It also needed to determine why the claim was procedurally defaulted. The Court did not speculate whether any state procedural rule could have justified the decision; it looked to the written opinion “unequivocally rest[ing] upon” the contemporaneous objection rule.310 The Court then remanded to the Ninth Circuit for consideration of whether Nunnemaker could demonstrate cause and prejudice for the failure to satisfy that particular state procedural rule.311 If the Court did not look through to the precise procedural reasons, there would be no discrete event for which the petitioner could show cause and prejudice.

Just as respecting the actual reasons given in a written opinion is paramount to the state judiciary function, so too should federal courts refrain from disregarding the likely meaning of a summary disposition. The chosen interpretive method for attributing content to a silent decision should reflect the intended meaning of the state court.312 From this perspective, Ylst is an excellent example of judicial comity. The Court established the look-through presumption to “most nearly reflect[ ] the role [summary dispositions] are ordinarily intended to play.”313 The presumption is rebuttable by strong evidence that a better explanation is available.314 The relevant question in this context is whether state courts that issue summary dispositions ordinarily intend to affirm the reasoning below.

To infer that state courts do not intend silent denials of nondiscretionary review to affirm the reasoning below, the Wilson majority relied on the practices of the Supreme Court and its own circuit.315 Two flaws afflict this approach. First, the “assumption that federal appellate practice should control” the understanding of state summary dispositions is an affront to comity.316 Second, this approach confuses a silent denial of nondiscretionary review with summary affirmances either after accepting nondiscretionary review or on direct appeal.317 These procedural differences prevent the easy transposition of inferences across these distinct contexts. The Wilson majority took a line of cases wholly out of context to argue against the look-through presumption.318 These cases involve the precedential effect of summary affirmances,319 which implicates prudential concerns regarding the use of summary dispositions as rules of decision in future cases.320 As § 2254(d) evaluates the content, and not the precedential effect, of a state decision for use in litigation on the same claim, the discussion of federal appellate practice is inapposite.

The debate is more properly focused on state practice. Opponents of the look-through presumption argue it makes “far more sense to assume that the [state appellate court] adhered to an established practice of summarily denying meritless claims rather than to presume” it “adopted wholesale the reasoning” of a lower court.321 By abstracting state practice to its most general terms, this argument serves only to confuse. Of course, the primary goal was to summarily deny an (at least seemingly) meritless claim, but the question is whether the issuance of a silent denial is more consistent with the court affirming the reasoning below or substituting its own reasoning. When a lower court renders a written opinion after discovery and an evidentiary hearing, this opinion is a prominent feature of the record.322

Currently, the crux of the debate turns on how one interprets reasoned denials that affirm the result but disagree with the reasoning below. Those opposing the look-through presumption argue that the fact that state courts sometimes issue reasoned denials is not sufficient to infer that silence is intended as adoption.323 Others would draw the exact opposite inference.324 The Georgia Supreme Court recently issued a decision that suggests that summary dispositions generally should be read to affirm the reasoning below.325 By issuing a summary disposition that explicitly declined to adopt the reasoning of the lower court, the court appears to accept the baseline presumption that a summary disposition affirms both result and reasoning. Some argue there exists no rationale for treating a summary disposition that explicitly refuses to affirm the reasoning differently from a standard summary disposition.326 This argument is clearly wrong: the objective of Ylst is the creation of an accurate and administrable presumption.327 If a court disclaims reliance on the reasoning below, the presumption is no longer accurate. But without such a disclaimer, federal courts have no basis for speculating in the face of a written opinion that has been summarily affirmed.

4. Opinion-writing requirements.

A closely related element of judicial comity is the Supreme Court’s prohibition on federal courts imposing opinion-writing requirements on state courts.328 The Wilson majority argued that the look-through presumption creates an opinion-writing requirement. The state appellate court would need to append “a statement of reasons” to prevent their decision from being interpreted as “a rubberstamp of the opinion below.”329 The reasons for the general prohibition are plentiful, such as heavy caseloads330 and preserving the case-law tradition,331 but the assumption that the look-through presumption imposes an opinion-writing requirement ignores the meaningful heterogeneity of state courts’ preferences. As the following analysis shows, determining the existence of an opinion-writing requirement is a complex inquiry depending on the practices and preferences of state courts. As discussed in the previous Section, state practices vary widely.332 Not surprisingly, there also exists disagreement over the preferences of state courts.

Professor Aziz Huq outlines two possible conceptual accounts for state courts’ preferences.333 The first is the “moral hazard theory of habeas.”334 The relevant conditions are that state courts derive negative utility from constitutional violations and that they are sensitive to changes in deference.335 In essence, when federal courts are more likely to act on constitutional violations, state courts are insured against their errors and will exercise less caution in avoiding them. The converse is likewise true: if federal courts extend more deference, state courts will be forced to work harder to avoid constitutional violations. Under these assumptions, refusing to apply the look-through presumption creates an incentive to write an opinion. Knowing a summary disposition lowers the chance relief will be granted, the state court would expend more energy checking for constitutional violations, and, if it agreed with the reasoned opinion below, the court would then need to affirmatively state so in order to lower the risk of a constitutional violation that evades federal review. Judge Guido Calabresi professes intuitions along these lines.336 He prefers allowing state courts to opt for less deference through silence, so “that their energy and resources [can be] better employed elsewhere.”337 Through a signaling mechanism to opt for less deference, states can “exercise [ ] control over their judicial resources which a true respect for state sovereignty requires.”338 If federal courts reject the look-through presumption, state courts with these preferences are required to write a reasoned denial affirming the reasoning below, which, contra Supreme Court direction, effectively imposes an opinion-writing requirement.

Huq’s second account involves a “‘sentinel effect,’ whereby the prospect of subsequent review induces greater care on the part of the front-line decision maker.”339 With a sentinel effect, the look-through presumption imposes the opposite incentives. If federal courts look through summary dispositions, state courts will be incentivized to state their disagreement with the reasoning in the lower-court opinion.340 Alternatively, without a look-through presumption, state courts will be encouraged to insulate their decisions by issuing summary dispositions.341 The Supreme Court found “no merit to the assertion that . . . [increased deference] . . . will encourage state courts to withhold explanations for their decisions,”342 but many observers beg to differ.343 Assuming the sentinel effect, the look-through presumption may apply pressure on state courts to write opinions, but the alternative—intentional insulation—flouts the state’s duty to vindicate constitutional rights in its postconviction system and undermines the public’s trust in the judiciary.344 The look-through presumption is the superior option under both accounts of state-court incentives.

C. Federalism

Federalism is frequently employed interchangeably with comity, but it is a distinct concept. Comity supports deference to another sovereign’s actions, but federalism allocates authority between state and national institutions.345 While comity most naturally exists between coequal sovereigns, federalism recognizes the supremacy of federal rights within the states.346 Thus, both states and individuals have a stake in federalism,347 which therefore has no inherent valence with respect to the availability of federal habeas relief.348 Indeed, the history of federal habeas has been one of expanding relief in the name of federalism. Federal habeas review of state prisoners’ claims originated as a tool “[t]o help superintend Reconstruction.”349 During the “due process revolution” of the 1960s and 1970s, expanded federal habeas review became necessary to protect constitutional rights against state noncompliance.350 At these points in time, the interests of federalism were advanced by increased federal intervention into the states’ criminal justice systems.351

Since Justice O’Connor’s opening line in Coleman v Thompson352 declared that “[t]his is a case about federalism,”353 the Court’s jurisprudence has demonstrated more solicitude for states’ interests than individuals’ rights.354 The states’ stake in federalism overlaps with the interests protected by comity, and, as shown by Parts IV.B.1 and IV.B.2, decreasing the availability of relief does not invariably serve the states’ interests.

Further, individual interests have not entirely disappeared from the scene.355 AEDPA’s deference regime necessitates the effective vindication of constitutional rights at the state level. The Court has demonstrated an interest in improving the quality of state postconviction review. Martinez (with its cousin Trevino v Thaler356 ) incentivizes states to appoint counsel to help petitioners develop claims of ineffective assistance of trial counsel.357 Montgomery v Louisiana358 requires that states give effect to new rules declared retroactive on state postconviction review.359 These cases illustrate a strong preference for deliberation over insulation. When a lower-court opinion is unreasonable under § 2254(d), a later silent denial does not inspire confidence that the state process is functioning properly.360 When unreasonable lower-court opinions are summarily affirmed, the state post­conviction review system lacks a feedback mechanism to re­inforce constitutional compliance.

Section 2254(d) does not demand willful blindness to unreasonable adjudications, nor does it require maximal deference on any ground imaginable. The objective is for states to fix constitutional errors in the first instance, not for federal courts to conceal them under the guise of hypothesizing. Doing so in the face of an unreasonable lower-court decision risks abdication by both sovereigns of the responsibility of adjudicating these claims. Federal courts should not unthinkingly press the cause of federalism against habeas petitioners without first considering the values of constitutional compliance and state deliberation.361

* * *

The text of § 2254(d) does not resolve the antecedent question of the domain of reasons available for deference. Looking through presumptively limits that domain to the reasons of the lower-court written opinion, while hypothesizing considers all possible reasons. This Comment has demonstrated that the look-through presumption is consistent with the Supreme Court’s post-Richter practice, AEDPA’s clear preference for evaluating reasoning, and the animating principles of federal habeas jurisprudence. The look-through presumption captures the meaning of the bulk of summary dispositions while retaining flexibility when the state court was unlikely to have endorsed the reasoning below.

The interest in finality, which rises monolithic from a final conviction, has no explanatory value in weighing the two approaches, while the concept of state-court fault explains why a subsequent summary disposition does not cure the fault emanating from an unreasonable written opinion.

Comity demands respect not only for the state’s interest in administering criminal justice, but also for its chosen structure for administering postconviction relief. Further, federal courts should give effect to the intended meaning of state-court opinions, not treat them as empty canvasses for their own reasoning. If state respondents can post hoc rationalize unreasonable decisions, the state forum’s position as the “main event” is undermined.

Lastly, invocations of federalism require sensitivity to the state-federal balance that seeks not only to further state control of core institutions but also to ensure the Constitution’s promise of individual liberty. To that end, the Court has required careful deliberation on state postconviction review. Values of constitutional compliance and norms against judicial insulation demand that federal courts not turn a blind eye to an unreasonable written opinion on the basis of a later silent decision.

Conclusion

For better or worse, finality, comity, and federalism loom over federal habeas jurisprudence. At present, these three principles operate primarily as abstract presumptions in favor of state respondents on ambiguous questions of AEDPA’s application. Until federal courts learn to wield these principles more skillfully, they are doomed to poorly reasoned approaches in which all three point invariably to maximal deference. A normative framework is not a compass. There is little value in utilizing a technique that points only in one direction despite its application to varied, nuanced, and difficult questions of interpretation. Finality, comity, and federalism each have their own constitutive elements that serve to limit their purposes. One must look to the structure of AEDPA and the Court’s own approach for guidance in resolving their conflicting aims.

By parsing these purposes in their concrete application to the issue at hand, federal courts can develop a more principled framework for resolving future ambiguities. This Comment attempts to reframe the discussion to this end and concludes that the look-through presumption best balances the bundle of normative commitments implicated by federal habeas review. Abstract invocations tantamount to “the prisoner always loses” are neither principled legally nor acceptable normatively. The hope is that close examination of finality, comity, and federalism will become the standard for hard questions of AEDPA interpretation.

By constructing AEDPA’s gates, Congress embodied its judgment on the availability of relief for state prisoners—in other words, which false negatives must go without remedy. Judges should not conjure Cerberus just because they think Congress did not go far enough in pursuing that end. When a state court silently blesses an unreasonable written opinion, the burden should not shift to the petitioner to rebut every hypothetical justification. Congress did not prescribe this outcome. Neither should the Court in Wilson v Sellers.

• 1Pub L No 104-132, 110 Stat 1214.
• 2Justice David Souter famously remarked that “in a world of silk purses and pigs’ ears, the Act is not a silk purse of the art of statutory drafting.” Lindh v Murphy, 521 US 320, 336 (1997). See also Lee Kovarsky, AEDPA’s Wrecks: Comity, Finality, and Federalism, 82 Tulane L Rev 443, 447 (2007) (describing the provisions of AEDPA as “hastily ratified and poorly cohered”).
• 3See, for example, Davis v Ayala, 135 S Ct 2187, 2197 (2015) (“For reasons of finality, comity, and federalism, habeas petitioners are not entitled to habeas relief based on trial error unless they can establish that it resulted in ‘actual prejudice.’”) (quotation marks omitted); McQuiggin v Perkins, 133 S Ct 1924, 1932 (2013) (“It would be passing strange to interpret a statute seeking to promote federalism and comity as requiring . . . .”).
• 4See Margaret A. Upshaw, Comment, The Unappealing State of Certificates of Appealability, 82 U Chi L Rev 1609, 1614–15 (2015) (“In light of these [drafting] shortcomings, any critical analysis drawing on statutory text and purpose must be approached with a significant measure of caution.”). This problem is not unique to AEDPA. For an articulation of this methodological defect in the administrative-law context, see Cass R. Sunstein and Adrian Vermuele, The Unbearable Rightness of Auer, 84 U Chi L Rev 297, 300 (2017) (“They invoke large abstractions . . . to resolve a concrete puzzle for which abstractions are either misplaced or unhelpful.”).
• 5In Greek mythology, Cerberus is a three-headed dog that guards the gates of Hades to keep the dead from escaping. See David Williams, Deformed Discourse: The Function of the Monster in Mediaeval Thought and Literature 128 (Exeter 1996). Some readers may be more familiar with “Fluffy,” who protects a trapdoor at Hogwarts. See J.K. Rowling, Harry Potter and the Sorcerer’s Stone 275–76 (Scholastic 1998).
• 6Wilson v Warden, Georgia Diagnostic Prison, 834 F3d 1227 (11th Cir 2016) (en banc), cert granted, 137 S Ct 1203 (2017).
• 7834 F3d 1227 (11th Cir 2016) (en banc). The en banc decision issued seventeen years after Wilson’s conviction was affirmed on direct review. See id at 1230. See also Joseph L. Hoffmann and Nancy J. King, Rethinking the Federal Role in State Criminal Justice, 84 NYU L Rev 791, 806–07 (2009) (discussing a “study [that] shows that the lag time from sentence to federal filing is over five years”).
• 8Petitioners name the custodian of their institution as the respondent to a writ of habeas corpus. See 28 USC § 2243. Lower courts frequently allow the use of the office as a placeholder—like “Warden, Georgia Diagnostic Prison”—but the Supreme Court may order that the official’s name (here, Sellers) be added. See US S Ct Rule 17(d).
• 9See 28 USC § 2254(b)(1).
• 1028 USC § 2241(c)(3).
• 11See Wilson, 834 F3d at 1230.
• 12See O’Sullivan v Boerckel, 526 US 838, 845 (1999) (“[S]tate prisoners must give the state courts one full opportunity to resolve any constitutional issues by invoking one complete round of the State’s established appellate review process.”).
• 13See Martinez v Ryan, 566 US 1, 13 (2012) (“Ineffective-assistance claims often depend on evidence outside the trial record. Direct appeals, without evidentiary hearings, may not be as effective as other proceedings for developing the factual basis for the claim.”).
• 14See Aziz Z. Huq, Habeas and the Roberts Court, 81 U Chi L Rev 519, 549 (2014) (describing these procedural barriers as the “Minoan labyrinth”); Stephen R. Reinhardt, The Demise of Habeas Corpus and the Rise of Qualified Immunity: The Court’s Ever Increasing Limitations on the Development and Enforcement of Constitutional Rights and Some Particularly Unfortunate Consequences, 113 Mich L Rev 1219, 1220 (2015) (lamenting the “twisted labyrinth of deliberately crafted legal obstacles”). See also Coleman v Thompson, 501 US 722, 759 (1991) (Blackmun dissenting) (“[T]he Court is creating a Byzantine morass of arbitrary, unnecessary, and unjustifiable impediments to the vindication of federal rights.”).
• 15Wilson, 834 F3d at 1230–31.
• 16See Ga Code Ann § 9–14–52(a).
• 17See Wilson, 834 F3d at 1232–33 (explaining that the Georgia Supreme Court will issue the CPC unless the appeal lacks “arguable merit”). AEDPA prescribes a similar limitation on appeals from adverse decisions by federal district courts, though it is phrased as a negative prohibition instead of an affirmative grant. See 28 USC § 2253(c)(2) (“A certificate of appealability may issue . . . only if the applicant has made a substantial showing of the denial of a constitutional right.”) (emphasis added).
• 18The Supreme Court usually does not grant certiorari just because a decision is incorrect. See US S Ct Rule 10 (“A petition for a writ of certiorari is rarely granted when the asserted error consists of erroneous factual findings or the misapplication of a properly stated rule of law.”).
• 19See Wilson, 834 F3d at 1234.
• 20See id at 1231. When referring to final determinations of a claim, courts use “summary disposition” and “summary order” interchangeably. Summary orders can also include nondispositive decisions, such as denials of motions to suppress evidence and motions for mistrial. In the interest of consistency and specificity, this Comment adheres to “summary disposition,” along with the more colloquial “silent denial.”
• 21See id.
• 22Greene v Fisher, 565 US 34, 40 (2011).
• 23See 28 USC § 2254(d).
• 24See Part I.B.1.
• 25See Brumfield v Cain, 135 S Ct 2269, 2276 (2015); Johnson v Tara Williams, 568 US 289, 297 n 1 (2013).
• 26See Harrington v Richter, 562 US 86, 100 (2011) (“[Section] 2254(d) does not require a state court to give reasons before its decision can be deemed to have been ‘adjudicated on the merits.’”).
• 27See Part I.B.2.
• 28See generally Wilson, 843 F3d 1227.
• 29See generally Cannedy v Adams, 706 F3d 1148 (9th Cir 2013) (“Cannedy I”); Grueninger v Director, Virginia Department of Corrections, 813 F3d 517 (4th Cir 2016).
• 30Judge Adalberto Jordan soon will discover the fate of his “prediction.” See Wilson, 834 F3d at 1242 (Jordan dissenting) (stylizing his opinion as a “prediction [ ] that the Supreme Court will . . . hold that the presumption [of looking through] in Ylst v. Nunnemaker governs”) (citation omitted).
• 31If the written opinion is reasonable, the petition will be denied under either approach: If the federal court looks through, the opinion will satisfy § 2254(d). If the federal court does not look through, the reasoning in the opinion is still a hypothetical basis supporting the later summary disposition.
• 32After bypassing § 2254(d), a petitioner technically must still prevail on de novo review. See note 79 and accompanying text. The process of demonstrating that a state decision was unreasonable, however, will almost always include the lesser showing that it was incorrect.
• 33Richter, 562 US at 101, quoting Yarborough v Alvarado, 541 US 652, 664 (2004). The Court’s retroactivity precedents use similar language to identify new rules. See O’Dell v Netherland, 521 US 151, 156 (1997) (“[W]e will not disturb a final state conviction or sentence unless it can be said that a state court . . . would have acted objectively unreasonably by not extending the relief later sought in federal court.”); Butler v McKellar, 494 US 407, 415 (1990) (stating that the outcome “was susceptible to debate among reasonable minds”).
• 34See Huq, 81 U Chi L Rev at 538 (cited in note 14) (describing how the Court rejected “less onerous” thresholds for relief in Richter).
• 35See Brandon L. Garrett and Lee Kovarsky, Federal Habeas Corpus: Executive Detention and Post-Conviction Litigation 302 (Foundation 2013) (stating that “the vast majority of state dispositions are in the form of summary orders”).
• 36Though the statute authorizing habeas review for federal prisoners, 28 USC § 2255, parallels that for state prisoners in most respects, the issue in this Comment is relevant only to state prisoners. Federal prisoners file in their court of conviction. 28 USC § 2255(a). The idea of a court deferring to its own decision is somewhat nonsensical (at least within the context of litigation on a single claim, as opposed to when applying norms of stare decisis). Strange too is the idea of a district court, on habeas review, overturning an authoritative decision issued by its circuit court of appeals on direct review. See Reed v Farley, 512 US 339, 358 (1994) (Scalia concurring in part and concurring in the judgment) (“[C]laims will ordinarily not be entertained under § 2255 that have already been rejected on direct review.”). Therefore, § 2255 lacks an analog to § 2254(d).
• 37See Justin F. Marceau, Challenging the Habeas Process Rather Than the Result, 69 Wash & Lee L Rev 85, 115 (2012) (“This question of how to deal with silent or summary state court decisions is not of interest only to academics or academically oriented judges; the deference owed to silent state court judgments is of immense practical importance.”).
• 38Black’s Law Dictionary 824 (Thomson Reuters 10th ed 2014).
• 39See Kovarsky, 82 Tulane L Rev at 446 n 9 (cited in note 2).
• 40See US Const Art I, § 9, cl 2 (“The Privilege of the Writ of Habeas Corpus shall not be suspended, unless when in Cases of Rebellion or Invasion the public Safety may require it.”).
• 41See Huq, 81 U Chi L Rev at 554 (cited in note 14) (“The consensus view today is that Congress can licitly withhold all postconviction review of state convictions, as it did until 1867.”). For a contrary view, see Jordan Steiker, Incorporating the Suspension Clause: Is There a Constitutional Right to Federal Habeas Corpus for State Prisoners?, 92 Mich L Rev 862, 873 (1994).
• 42Judiciary Act of 1789 § 14, 1 Stat 73, 81–82.
• 43Habeas Corpus Act of 1867 § 1, 14 Stat 385, 386. Prior to 1867, Congress had authorized federal courts to entertain habeas petitions for limited subsets of state prisoners, usually in furtherance of specifically federal interests. See, for example, Force Act of 1833 § 7, 4 Stat 632, 634–35 (extending jurisdiction to prisoners in state custody “for any act done, or omitted to be done, in pursuance of a law of the United States, or any order, process, or decree, of any judge or court thereof”).
• 44See, for example, Ex parte Siebold, 100 US 371, 375 (1879) (“[T]he general rule is, that a conviction and sentence by a court of competent jurisdiction is lawful cause of imprisonment, and no relief can be given by habeas corpus.”); Frank v Magnum, 237 US 309, 327 (1915).
• 45Hoffmann and King, 84 NYU L Rev at 801 (cited in note 7).
• 46344 US 443 (1953).
• 47Id at 458–59 (affirming “the power of the [federal] [c]ourt to reexamine federal constitutional issues even after trial and [direct] review”). Some consider Brown an “anticlimax,” doing no more than “nicely catalogu[ing] the governing habeas corpus principles” at the time. James S. Liebman, Apocalypse Next Time?: The Anachronistic Attack on Habeas Corpus/Direct Review Parity, 92 Colum L Rev 1997, 2083 (1992).
• 48See generally Barry Friedman, Failed Enterprise: The Supreme Court’s Habeas Reform, 83 Cal L Rev 485 (1995).
• 49See Kovarsky, 82 Tulane L Rev at 447, 463 (cited in note 2) (identifying the “political climate that allowed Republicans to append habeas reform to antiterrorism legislation”).
• 50See text accompanying notes 67–70.
• 51AEDPA § 104, 110 Stat at 1219, 28 USC § 2254(d).
• 52Matthew Seligman, Note, Harrington’s Wake: Unanswered Questions on AEDPA’s Application to Summary Dispositions, 64 Stan L Rev 469, 470 (2012).
• 53505 US 277 (1992).
• 54Id at 291.
• 55Id at 305 (O’Connor concurring in the judgment).
• 56Id (O’Connor concurring in the judgment). For extensive discussion of the historical accuracy of the dueling opinions in West, see generally Liebman, 92 Colum L Rev 1997 (cited in note 47).
• 57See Mark Tushnet and Larry Yackle, Symbolic Statutes and Real Laws: The Pathologies of the Antiterrorism and Effective Death Penalty Act and the Prison Litigation Reform Act, 47 Duke L J 1, 44, 47 (1997) (predicting “marginal results” that “change the flavor rather little” relative to preexisting practice).
• 58See id at 45.
• 59See id.
• 60529 US 362 (2000).
• 61Justice John Paul Stevens, writing for a noncontrolling plurality on the issue, would have adopted the appellate-review interpretation. See id at 389 (Stevens) (plurality) (“In sum, the statute directs federal courts to attend to every state-court judgment with utmost care, but it does not require them to defer to the opinion of every reasonable state-court judge on the content of federal law.”).
• 62Id at 410.
• 63Id at 411. O’Connor did not claim that Congress knowingly adopted Thomas’s rule. See id (“In any event, whether Congress intended to codify the standard of review suggested by Justice Thomas in [West] is beside the point.”).
• 64Terry Williams, 529 US at 405–06.
• 65Id at 413.
• 66Id at 409. The “reasonable jurist” standard “may be misleading,” so a federal habeas court should avoid a subjective inquiry that “rest[s] its determination [ ] on the simple fact that at least one . . . jurist[ ]” applied federal law in such a manner in the petitioner’s case. Id at 409–10.
• 67See 28 USC § 2254(b)–(c) (codifying exhaustion requirements); Rose v Lundy, 455 US 509, 522 (1982) (prescribing a “total exhaustion rule” whereby petitions containing a mix of exhausted and unexhausted claims are dismissed in their entirety).
• 68Procedural default is a judge-made doctrine that coexists with its statutory brethren. See Wainwright v Sykes, 433 US 72, 81 (1977). If the state decision rested on an adequate and independent state-law ground—such as res judicata or a contemporaneous objection rule—the federal court cannot review the claim unless a petitioner meets the cause-and-prejudice standard or shows a miscarriage of justice. See id at 87, 91.
• 69See 28 USC § 2244(b).
• 70See 28 USC § 2244(d).
• 71For a visual representation of the procedural gates, see Lee Kovarsky, The Habeas Optimist, 81 U Chi L Rev Dialogue 108, 104 (2014), archived at http://perma.cc/634V-YF2E (Figure 1).
• 72See, for example, Johnson v Tara Williams, 568 US 289, 297 (2013); Marceau, 69 Wash & Lee L Rev at 108 (cited in note 37).
• 73See Kovarsky, 81 U Chi L Rev Dialogue at 105–06 (cited in note 71). See also Harrington v Richter, 562 US 86, 100 (2011) (describing § 2254(d) as a “relitigation bar”).
• 74563 US 170 (2011).
• 75Id at 181.
• 76565 US 34 (2011).
• 77Id at 39 (holding that “clearly established Federal law” is judged from the perspective of the last state court to adjudicate the merits).
• 78Kovarsky, 81 U Chi L Rev Dialogue at 103 (cited in note 71).
• 79See id at 106 (identifying the “single terminal inquiry” after passing all procedural gates as “the merits of the constitutional claim”). See also, for example, Cannedy v Adams, 706 F3d 1148, 1166 (9th Cir 2013) (“Cannedy I”) (“Having concluded that the state court’s decision was unreasonable, we review the substantive constitutionality of the state custody de novo.”) (quotation marks omitted). Of course, in almost all cases, demonstrating that the state decision is unreasonable will entail prevailing on the merits. But if the state decision relies on a constitutional rule “contrary to” Supreme Court precedent, the petitioner may not necessarily prevail on the merits in the federal habeas court. See text accompanying note 64.
• 80Kovarsky, 81 U Chi L Rev Dialogue at 106 (cited in note 71) (emphasis omitted). Commentators have analogized § 2254(d)(1) review of state-court fault to the abuse-of-discretion standard in administrative law and qualified immunity for constitutional torts. See Allan Ides, Habeas Standards of Review Under 28 U.S.C. § 2254(d)(1): A Commentary on Statutory Text and Supreme Court Precedent, 60 Wash & Lee L Rev 677, 691–92 n 27 (2003) (abuse of discretion); Huq, 81 U Chi L Rev at 583–84 (cited in note 14) (qualified immunity).
• 81Seligman, Note, 64 Stan L Rev at 471 (cited in note 52).
• 82Id at 474.
• 83See id at 506 (remarking that, “in some jurisdictions, [the problem of interpreting a summary disposition] arises in virtually all federal habeas petitions”).
• 84See id at 503–06 (compiling summary disposition rates from select states).
• 85501 US 797 (1991).
• 86384 US 436 (1966).
• 87Ylst, 501 US at 799.
• 88Id (“The sole basis for its rejection of the Miranda claim was the state procedural rule that an objection based upon a Miranda violation cannot be raised for the first time on appeal.”) (quotation marks omitted).
• 89Id at 799–800.
• 90See note 68.
• 91See Ylst, 501 US at 801.
• 92Id.
• 93Id.
• 94Id at 803 (emphasis added).
• 95Ylst, 501 US at 803.
• 96Id at 804. In adopting the look-through presumption, the Ylst Court departed from its rule in Michigan v Long, 463 US 1032 (1983). On direct review of a state-court judgment, the Court searches only the four corners of the opinion for the existence of an adequate and independent state-law ground. Id at 1040–41 (explaining that “when the adequacy and independence of any possible state-law ground is not clear from the face of the opinion,” the state-court decision is presumed to rest on federal grounds). The absence of an adequate and independent state-law ground is jurisdictional on direct review, see Fox Film Corp v Muller, 296 US 207, 210–11 (1935), while procedural default is nonjurisdictional for federal habeas courts, see Trest v Cain, 522 US 87, 89 (1997). If Ylst adopted Long’s rule, procedural bars would routinely be vitiated by summary dispositions—the four corners of silent decisions, unsurprisingly, rarely contain clear statements of state-law grounds. Long increased the Court’s power to hear (and reverse) state-court judgments (such as pro-defendant criminal-procedure rulings). Taken with Ylst’s concomitant decrease in the lower federal courts’ ability to review the convictions of state prisoners, some contemporary scholars perceived the Court as using procedural arcana to further ideological ends. See, for example, Christopher Slobogin, Having It Both Ways: Proof That the U.S. Supreme Court Is “Unfairly” Prosecution-Oriented, 48 Fla L Rev 743, 755–57 (1996). A cynic might predict that the look-through presumption became less attractive once § 2254(d) replaced procedural default as the biggest hurdle to merits review. See text accompanying notes 71–80. If insulating merits-based reasons is a better tool for denying petitions than uncovering procedural ones—the intuition goes—a Court with such preferences may more easily discard logical applications and extensions of the look-through presumption. This Comment, however, takes Ylst at face value and is content to leave legal-realist speculation for another day.
• 97Ylst, 501 US at 804.
• 98See id at 806.
• 99Id at 804.
• 100See, for example, Kernan v Hinojosa, 136 S Ct 1603, 1605–06 (2016) (per curiam); Richter, 562 US at 99–100.
• 101135 S Ct 2269 (2015).
• 102See id at 2276. See also Tara Williams, 568 US at 297 n 1.
• 103For more discussion on the distinctions between types of review and bases of decision, see text accompanying notes 128–36.
• 104562 US 86 (2011).
• 105Id at 94–96.
• 106Id at 96. The decision in its entirety read: “Petition for Writ of Habeas Corpus is DENIED.” Respondent’s Brief on the Merits, Harrington v Richter, Docket No 09-587, *11–12 (US filed July 9, 2010) (available on Westlaw at 2010 WL 2770109).
• 107Richter, 562 US at 97.
• 108Id.
• 109Id.
• 110See Seligman, Note, 64 Stan L Rev at 471 (cited in note 52) (discussing the range of solutions “from great deference to de novo review”).
• 111See, for example, Brittany Glidden, Note, When the State Is Silent: An Analysis of AEDPA’s Adjudication Requirement, 27 NYU Rev L & Soc Change 177, 205 (2002) (deeming it “uncertain” whether “an actual state adjudication of a federal right” occurred); Robert D. Sloane, Comment, AEDPA’s “Adjudication on the Merits” Requirement: Collateral Review, Federalism, and Comity, 78 St John’s L Rev 615, 619–20 (2004) (concluding that summary dispositions should not be treated as “adjudications on the merits”) (alteration omitted).
• 112See, for example, Claudia Wilner, Note, “We Would Not Defer to That Which Did Not Exist”: AEDPA Meets the Silent State Court Opinion, 77 NYU L Rev 1442, 1473 (2002) (interpreting “unreasonable application” in § 2254(d) to encompass decisions that lack reasoning).
• 113Richter, 562 US at 98. The Court also established a presumption that summary dispositions are decisions on the merits. See id at 99 (“When a federal claim has been presented to a state court and the state court has denied relief, it may be presumed that the state court adjudicated the claim on the merits in the absence of any indication or state-law procedural principles to the contrary.”).
• 114Id at 98.
• 115Id.
• 116Id at 101.
• 117Richter, 562 US at 102 (emphasis added).
• 118See id at 103 (identifying “repose for concluded litigation,” state courts’ position as the “principal forum for asserting constitutional challenges to state convictions,” and a reluctance to “intrude[ ] on state sovereignty” as reasons for the approach).
• 119Id at 100. Apparently the Ylst Court engaged in “pure speculation.” See Ylst, 501 US at 803 (“The problem we face arises, of course, because many formulary orders are not meant to convey anything as to the reason for the decision.”).
• 120Richter, 562 US at 100.
• 121Id at 102.
• 122Id at 113.
• 123See id at 99.
• 124Richter, 562 US at 99.
• 125Id.
• 126568 US 289 (2013).
• 127Id at 300.
• 128See text accompanying note 81.
• 129See Dennis v Secretary, Pennsylvania Department of Corrections, 834 F3d 263, 372 n 4 (3d Cir 2016) (en banc) (Hardiman dissenting) (collecting cases). These judges decry the asymmetry in which poorly reasoned decisions receive less deference than silent ones. See, for example, Mann v Ryan, 774 F3d 1203, 1224–25 (9th Cir 2014) (Kozinski concurring in part and dissenting in part). Their preferred means of reinstating symmetry, apparently, would be to make all petitioners worse off, a maneuver called “leveling down” in the equal-protection context. See generally Deborah L. Brake, When Equality Leaves Everyone Worse Off: The Problem of Leveling Down in Equality Law, 46 Wm & Mary L Rev 513 (2004). See also Ian Samuel, Morales-Santana and the “Mean Remedy” (PrawfsBlawg, June 12, 2017), archived at http://perma.cc/W6YS-HCBD.
• 130See, for example, Noam Biale, Beyond a Reasonable Disagreement: Judging Habeas Corpus, 83 U Cin L Rev 1337, 1340–41 (2015) (“The subjectivism now creeping into the habeas opinions of numerous circuit judges applying Richter . . . cannot be squared with a proper understanding of the Supreme Court’s habeas jurisprudence.”); Huq, 81 U Chi L Rev at 538 n 88 (cited in note 14) (“It is tolerably clear that Richter has not displaced the Terry Williams rule with respect to the ‘contrary to’ element of § 2254(d)(2).”), citing Metrish v Lancaster, 133 S Ct 1781, 1787 n 2 (2013).
• 131See Williams v Cavazos, 646 F3d 626, 636 (9th Cir 2011), revd on other grounds, Tara Williams, 568 US 289.
• 132Brumfield, 135 S Ct at 2275.
• 133Id at 2276.
• 134Brief of Adam K. Mortara, Court-Appointed Amicus Curiae Below, as Amicus Curiae in Support of Respondent, Wilson v Sellers, Docket No 16-6855, *4 (US filed Jan 5, 2017) (available on Westlaw at 2017 WL 83633) (“Mortara Brief”).
• 135For an argument that the level of deference given to a summary disposition should depend on the state’s deliberative processes, see Seligman, Note, 64 Stan L Rev at 485–88 (cited in note 52).
• 136See Eliza Beeney, Note, Why Silence Shouldn’t Speak So Loudly: Wiggins in a Post-Richter World, 101 Cornell L Rev 1321, 1337 (2016) (“However, in a post-Richter world, it has become clear that Ylst applies to summary denials and that federal habeas courts can ‘look through’ summary denials to the reasons that a lower state court has given for rejecting a prisoner’s claim.”); Seligman, Note, 64 Stan L Rev at 484 (cited in note 52) (explaining that evaluating a summary disposition “does not pose significant problems” when a federal court can look through to a lower court’s written opinion).
• 137See Ylst, 501 US at 797.
• 138706 F3d 1148 (9th Cir 2013).
• 139Id at 1153.
• 140Id at 1154.
• 141Id at 1154–55.
• 142Cannedy I, 706 F3d at 1155. To establish ineffective assistance, a petitioner must prevail on both prongs of the test established by Strickland v Washington, 466 US 668, 687 (1984).
• 143Cannedy I, 706 F3d at 1155.
• 144Id at 1156.
• 145Id.
• 146Id at 1157.
• 147Cannedy I, 706 F3d at 1158 (collecting examples from other circuits).
• 148Id.
• 149Id at 1166–67 (Kleinfeld dissenting). See also Cannedy v Adams, 733 F3d 794, 795 (9th Cir 2013) (“Cannedy II”) (O’Scannlain dissenting from denial of rehearing en banc) (“[T]he court regrettably disregards explicit guidance from the Supreme Court.”).
• 150Cannedy I, 706 F3d at 1167 (Kleinfeld dissenting). For the full quote from Richter, see text accompanying note 117.
• 151Cannedy I, 706 F3d at 1167 (Kleinfeld dissenting).
• 152Id at 1156 & n 3.
• 153See id at 1169 (Kleinfeld dissenting) (decrying the majority’s “contortion of looking through . . . to the prior California Court of Appeal decision, [to] deem[ ] it unreasonable based on what Cannedy never submitted to the Court of Appeal”).
• 154Cannedy II, 733 F3d at 801 (O’Scannlain dissenting from denial of rehearing en banc), citing Pinholster, 563 US at 183 n 3.
• 155Cannedy II, 733 F3d at 802 (O’Scannlain dissenting from denial of rehearing en banc), quoting Cannedy I, 706 F3d at 1159 n 5. This response is somewhat unconvincing, as “federal courts have no authority to impose mandatory opinion-writing standards on state courts.” Tara Williams, 568 US at 300. Instead, one could argue that the state court would have provided different reasoning had its decision rested on alternative grounds. The latter inquiry drives to the heart of what the decision meant, not what a federal court requires of a state court to discharge its burden under § 2254(d). When possible, federal courts should give effect to the intended meaning of a state-court decision. For a discussion of the competing inferences, see Part IV.B.4.
• 156See text accompanying notes 97–99.
• 157Cannedy II, 733 F3d at 801–02 (O’Scannlain dissenting from denial of rehearing en banc) (“This approach satisfies both the rule in Ylst and the reasoning in Pinholster, ensuring that the two decisions can coexist harmoniously rather than standing in tension.”).
• 158O’Scannlain advocates a stronger version of this argument, such that the presumption is rebutted whenever there is new evidence on the record. See id at 797 (O’Scannlain dissenting from denial of rehearing en banc).
• 159See id at 802 (O’Scannlain dissenting from denial of rehearing en banc).
• 160813 F3d 517 (4th Cir 2016).
• 161Id at 520.
• 162Id at 520, 522.
• 163Id at 522.
• 164Grueninger, 814 F3d at 523.
• 165Id.
• 166Id at 525.
• 167Id, citing Richter, 562 US at 96–97.
• 168Grueninger, 814 F3d at 525.
• 169Id at 526.
• 170Id.
• 171For a refresher, see text accompanying notes 131–35.
• 172Wilson was introduced at the beginning of this Comment. See text accompanying notes 6–20.
• 173Wilson, 834 F3d at 1230.
• 174Id.
• 175Id at 1230–31.
• 176Id at 1232. See also text accompanying notes 16–19.
• 177Wilson, 834 F3d at 1231.
• 178Id.
• 179See text accompanying notes 145–46, 166–67.
• 180Wilson, 834 F3d at 1235 (“Nothing in . . . Richter suggests that its reasoning is limited to the narrow subset of habeas petitions where there is no reasoned decision from any state court.”).
• 181Id at 1236.
• 182Tara Williams, 568 US at 300.
• 183Wilson, 834 F3d at 1238.
• 184Id at 1240.
• 185See text accompanying note 99.
• 186Wilson, 324 F3d at 1240.
• 187Coleman v Thompson, 501 US 722, 739 (1991). See also Foster v Chatman, 136 S Ct 1737, 1746 n 3 (2016) (“There would be no way to know [the meaning of the decision], of course, from the face of the Georgia Supreme Court’s summary order.”).
• 188See Wilson, 834 F3d at 1268 n 23 (Pryor dissenting). See also Pinholster, 563 US at 182 (“[T]he record under review is limited to the record in existence at that same time—i.e., the record before the state court.”).
• 189See Wilson, 834 F3d at 1236.
• 19028 USC § 2254(d)(1).
• 191For the text of § 2254(d), see text accompanying note 51.
• 192Such a scheme penalizes states whose courts issue opinions. See note 341 and accompanying text.
• 193See text accompanying notes 62–66.
• 194See Part IV.B.3.
• 195See Wilson, 834 F3d at 1251–54 (Pryor dissenting) (arguing that the mode of analysis depends on the existence of a written opinion).
• 196Id at 1242–43 (Jordan dissenting).
• 197562 US 115 (2011).
• 198Id at 119–20.
• 199See Moore v Palmeteer, 26 P3d 191 (Or App 2001) (table).
• 200Moore, 562 US at 132.
• 201See id.
• 202See note 31.
• 203136 S Ct 1737 (2016).
• 204Id at 1745–47.
• 205Id at 1764–65 (Thomas dissenting).
• 206Lawrence v Florida, 549 US 327, 335 (2007) (quotation marks omitted).
• 207Foster, 136 S Ct at 1761 (Alito concurring in the judgment).
• 208See Wilson, 834 F3d at 1263–64 & n 17 (Pryor dissenting).
• 209Tara Williams, 568 US at 293 (stating that the result “follows logically from [Richter]”).
• 210Id at 298.
• 211566 US 156 (2012).
• 212Id at 173. The dissent would have applied Richter deference. See id at 183 (Scalia dissenting).
• 213Brumfield, 135 S Ct at 2282–83 (emphasis added), quoting Richter, 562 US at 98.
• 214135 S Ct 2126 (2015).
• 215Id at 2127 (Ginsburg concurring in denial of certiorari), citing Hittson v GDCP Warden, 759 F3d 1210, 1232 n 25 (11th Cir 2014).
• 216Hittson, 135 S Ct at 2128 (Ginsburg concurring in denial of certiorari).
• 217Id at 2127 (Ginsburg concurring in denial of certiorari).
• 218Id at 2126 (Ginsburg concurring in denial of certiorari).
• 219Id at 2127 (Ginsburg concurring in denial of certiorari).
• 220Hittson, 135 S Ct at 2127 (Ginsburg concurring in denial of certiorari).
• 221Id at 2127 (Ginsburg concurring in denial of certiorari).
• 222Id at 2127–28 (Ginsburg concurring in denial of certiorari) (alterations in original), quoting Richter, 562 US at 102.
• 223See Part III.A.
• 224See Part III.B.
• 225Michael Williams v Taylor, 529 US 420, 436 (2000).
• 226Kovarsky, 82 Tulane L Rev at 444–45 & n 5 (cited in note 2) (collecting recent Supreme Court cases endorsing this “mood”).
• 227See, for example, id at 445 (“Given what we know about AEDPA’s legislative history, there is little support for the argument that courts should interpret AEDPA’s ambiguities with any particular purposes in mind.”).
• 228See, for example, Woods v Donald, 135 S Ct 1372, 1376 (2015) (“Adherence to these principles serves important interests of federalism and comity.”); Pinholster, 563 US at 185 (invoking “AEDPA’s goal of promoting comity, finality, and federalism”) (quotation marks omitted), quoting Jimenez v Quarterman, 555 US 113, 121 (2009); Panetti v Quarterman, 551 US 930, 947 (2007) (discussing how AEDPA is “implemented to further the principles of comity, finality, and federalism”).
• 229See, for example, Ex parte Bollman, 8 US (4 Cranch) 75, 92 (1807) (“[T]here may properly be a comity observed which would prevent them from attempting to interfere with the decisions of each other.”); Ex parte Royall, 117 US 241, 251 (1886) (recognizing “the fact that the public good requires that those relations [between federal and state courts] be not disturbed by unnecessary conflict”); Frank v Magnum, 237 US 309, 334 (1915) (committing to consider the entire state process because to do otherwise would be “to disregard comity”); Rose v Lundy, 455 US 509, 522 (1982) (adopting a total exhaustion rule “because [it] promotes comity”); Stringer v Black, 503 US 222, 228 (1992) (“The interests in finality, predictability, and comity underlying our new rule jurisprudence . . . .”); Davila v Davis, 137 S Ct 2058, 2064 (2017) (“The procedural default doctrine thus advances the same comity, finality, and federalism interests advanced by the exhaustion doctrine.”).
• 230See Kovarsky, 82 Tulane L Rev at 455 (cited in note 2) (“Where one interest obviously trades off with another, a specific textual formulation was as likely to represent a limit on an animating purpose as it is to represent an endorsement of it.”).
• 231Id at 470.
• 232Id at 470–71 (“AEDPA’s text can tell judges which purposes compete, but not which ones win.”).
• 233See id at 471 (“By citing to generalized purposes, courts decide the relative intensities of competing purposes where the legislature has declined to do so.”).
• 234American Express Co v Italian Colors Restaurant, 133 S Ct 2304, 2309 (2013), quoting Rodriguez v United States, 480 US 522, 525–26 (1987) (per curiam).
• 235See, for example, Richter, 562 US at 103 (“[Federal habeas review] ‘disturbs the State’s significant interest in repose for concluded litigation[ ] [and] denies society the right to punish some admitted offenders.’”), quoting Harris v Reed, 489 US 255, 282 (1989) (Kennedy dissenting).
• 236See Kovarsky, 82 Tulane L Rev at 503 (cited in note 2) (identifying Bator as “the intellectual patriarch of modern habeas reform”).
• 237Paul M. Bator, Finality in Criminal Law and Federal Habeas Corpus for State Prisoners, 76 Harv L Rev 441, 443 (1963).
• 238See id at 451–52.
• 239Id at 447 (warning that “there [would] be no escape from a literally endless re­litigation of the merits because the possibility of mistake always exists”).
• 240Chief Justice William Rehnquist, then a law clerk to Justice Robert Jackson, encouraged his justice’s famous concurrence in Brown with an argument along these lines. See Garrett and Kovarsky, Federal Habeas Corpus at 131 (cited in note 35) (discussing Rehnquist’s memorandum entitled “Habeas Corpus, Then and Now, Or, ‘If I Can Just Find the Right Judge, Over These Prison Walls I Shall Fly’”).
• 241See id at 456 (advocating “a full and fair opportunity” to present the claim as the prerequisite to preclusion in a federal forum). For modern defenses of a process-oriented focus for § 2254(d), see generally Biale, 83 U Cin L Rev 1337 (cited in note 130); Marceau, 69 Wash & Lee L Rev 85 (cited in note 37).
• 242Huq, 81 U Chi L Rev at 539–40 (cited in note 14).
• 243See Richter, 562 US at 102–03 (describing federal habeas review as “a guard against extreme malfunctions in the state criminal justice systems, not a substitute for ordinary error correction through appeal”) (quotation marks omitted); Lockyer v Andrade, 538 US 63, 75 (2003) (chastising the circuit court for “conflating error (even clear error) with unreasonableness”). See also Kovarsky, 81 U Chi L Rev Dialogue at 110 (cited in note 71) (describing § 2254(d) as a “fault-identifying rule rather than an error-identifying one”). For a discussion of the difference between “fault” and “error,” see text accompanying notes 75–80.
• 244See Part I.A.
• 245See Brecht v Abrahamson, 507 US 619, 635 (1993) (“distinguishing between direct and collateral review” by referencing “the [s]tate’s interest in the finality of convictions that have survived direct review”).
• 246A plausible habeas regime could predicate relitigation of federal claims on any of these normative values (or a combination thereof). In fact, one need not look far for a habeas doctrine sensitive to fault, error, and innocence. The law of procedural default governs the circumstances under which a federal court can hear a claim in the first instance when the state court will not for procedural reasons. See note 68. To circumvent the procedural bar, a petitioner can either show cause (a lack of fault) and prejudice (error), Wainwright v Sykes, 433 US 72, 87 (1977), or “establish sufficient doubt about his guilt” (innocence), Schlup v Delo, 513 US 298, 316 (1995).
• 247Section 2254(d)’s trigger for merits review in federal court is a showing of state-court fault. See text accompanying notes 78–80.
• 248See text accompanying notes 131–33.
• 249See text accompanying note 189.
• 250See Ryan Bubb and Richard H. Pildes, How Behavioral Economics Trims Its Sails and Why, 127 Harv L Rev 1593, 1619 n 82 (2014).
• 251See, for example, Stephen Yelderman, The Value of Accuracy in the Patent System, 84 U Chi L Rev 1217, 1224–26 (2017); Daniel Epps, The Consequences of Error in Criminal Justice, 128 Harv L Rev 1065, 1068 (2015).
• 252See text accompanying notes 67–73.
• 253See Mitchell N. Berman, Constitutional Decision Rules, 90 Va L Rev 1, 93 (2004) (“The most obvious factor that a decision-rule-maker should consider, then, is how best to minimize adjudicatory errors—i.e., the sum of false positives and false negatives.”). The risk of factual error prior to conviction (in other words, an innocent person being found guilty) is treated in a strikingly different manner from the risk of unconstitutional detention following a conviction. The entrenched view is Blackstone’s principle, that it is “better that ten guilty persons escape, than that one innocent suffer.” Epps, 128 Harv L Rev at 1067 (cited in note 251) (alteration omitted). Behind the Rawlsian veil of ignorance, one reasonably might accept the procedural asymmetries that criminal defendants enjoy against prosecutors—like an elevated burden of proof and the state’s inability to appeal an acquittal—in exchange for the unforgiving gates that AEDPA erects on postconviction review. Blackstone’s ratio, however, may insufficiently take account of its potentially negative impact on criminal defendants. See generally id (exploring a dynamic model of error costs). One such cost is the substitution from the jury trial to a less procedurally rigorous alternative (plea bargaining). See id at 1144 (discussing factors that “might unduly encourage innocent defendants to plead guilty notwithstanding the prospect of a trial conducted under defendant-friendly procedural rules”). Lastly, if the public and the judiciary believe that all those that are convicted are almost certainly guilty in light of the formal safeguards, they likely will exhibit less solicitude for prisoners’ constitutional rights.
• 254Many legal regimes accept higher rates of errors because the costs of administering a more accurate rule are not worth the corresponding gains from decreasing errors. See Cass R. Sunstein, The Supreme Court 1995 Term—Foreword: Leaving Things Undecided, 110 Harv L Rev 4, 16 (1996) (“As a first approximation, we might try to systematize the inquiry . . . in the following way: good judges try to minimize the sum of decision costs and error costs.”). Federal habeas corpus plausibly could be at home under this definition, but considerations other than accuracy or cost likely explain AEDPA’s strong preference for false negatives. See Adam M. Samaha, Undue Process, 59 Stan L Rev 601, 639 & n 150, 664 n 244 (2006) (finding an uneasy fit between habeas rules and the decision-cost rationale).
• 255See, for example, Huq, 81 U Chi L Rev at 582–93 (cited in note 14) (evaluating “[f]ault as lodestar”); Erica Hashimoto, Reclaiming the Equitable Heritage of Habeas, 108 Nw U L Rev 139, 150–63 (2014) (detailing the “equitable roots” of most of the habeas gates).
• 256Huq, 81 U Chi L Rev at 581 (cited in note 14).
• 257See text accompanying notes 78–80.
• 258See Herrera v Collins, 506 US 390, 399–400 (1993).
• 25928 USC § 2254(b)(1)(B)(i).
• 260See, for example, Strickler v Greene, 527 US 263, 282 (1998) (finding that state suppression of material documents constitutes cause).
• 261See Coleman v Thompson, 501 US 722, 753–54 (1991) (rejecting the claim that attorney inadvertence can be cause for bypassing procedural default). The standard rationale is that the mistakes of the agent (the lawyer) are ascribed to the principal (the petitioner). See Murray v Carrier, 477 US 478, 488 (1986) (“[W]e discern no inequity in requiring him to bear the risk of attorney error that results in a procedural default.”). For an argument that a structural approach to procedural adequacy is preferable to the cause-and-prejudice standard, in part for focusing on state fault, see Eve Brensike Primus, Federal Review of State Criminal Convictions: A Structural Approach to Adequacy Doctrine, 115 Mich L Rev *41–53 (forthcoming 2017), archived at http://perma.cc/P64R-57B6.
• 262See Carrier, 477 US at 488.
• 263566 US 1, 9 (2012) (“Inadequate assistance of counsel at initial-review collateral proceedings may establish cause for a prisoner’s procedural default of a claim of ineffective assistance at trial.”).
• 264See Huq, 81 U Chi L Rev at 584 n 283 (cited in note 14) (“Federal courts have systematically ignored the paradox that results from attributing state-funded lawyers’ errors to petitioners when those errors are more plausibly traced back to (under-)funding decisions by state legislatures.”); Primus, 115 Mich L Rev at *4 (cited in note 261) (“In short, the system engenders widespread ineffectiveness of defense representation by refusing to fund indigent defense adequately, and then it prevents courts from redressing the resulting constitutional violations by creating procedural barriers to reviewing claims of ineffective assistance.”).
• 265489 US 288 (1989).
• 266See id at 310. Teague prevents a petitioner from relying on a “new rule”—one announced after his conviction was final on direct review—on federal habeas review unless the rule meets an exception. See id. The Court identified finality as a major factor in its decision. See id at 308–09 (“[W]e have recognized that interests of comity and finality must also be considered in determining the proper scope of habeas review.”). Teague, which predicates the availability of a new rule on temporal sequencing wholly outside the petitioner’s control, has been identified as an “equitable outlier.” Hashimoto, 108 Nw U L Rev at 163, 170–72 (cited in note 255).
• 267See Danforth v Minnesota, 552 US 264, 281–82 (2008). This is despite the fact that Teague is nominally grounded in finality and treating “similarly situated” petitioners equally. Teague, 489 US at 316 (stating that the retroactivity bar “avoids the inequity resulting from the uneven application of new rules to similarly situated defendants”). If this—rather than a desire to refrain from intrusion on the states except when they are at fault—were the true rationale, the dissent’s argument would have carried the day. See Danforth, 552 US at 300–01 (Roberts dissenting).
• 268See Montgomery v Louisiana, 136 S Ct 718, 731–32 (2016).
• 269See Ylst, 501 US at 804.
• 270See Cannedy v Adams, 733 F3d 794, 800–01 (9th Cir 2013) (“Cannedy II”) (O’Scannlain dissenting from denial of rehearing en banc) (arguing that federal courts should not look through if it requires “presum[ing] that the [state] court ignored or overlooked new evidence”).
• 271See Mortara Brief at *19 n 10 (cited in note 134). See also Wilson, 834 F3d at 1238 (“[E]ven when the opinion of a lower state court contains flawed reasoning, the Act requires that we give the last state court to adjudicate the prisoner’s claim on the merits ‘the benefit of the doubt.’”), quoting Renico v Lett, 559 US 766, 773 (2010).
• 272See text accompanying note 250.
• 273Richter, 562 US at 102.
• 274Ylst, 501 US at 804.
• 275See Wilson, 834 F3d at 1247 (Pryor dissenting) (“[R]ejecting a look-through presumption . . . places a far heavier burden on habeas petitioners than [AEDPA] requires.”).
• 276Kovarsky, 82 Tulane L Rev at 455 (cited in note 2).
• 277For an oft-quoted definition of international comity, see Hilton v Guyot, 159 US 113, 163–64 (1895):

  “Comity,” in the legal sense, is neither a matter of absolute obligation, on the one hand, nor of mere courtesy and good will, upon the other. But it is the recognition which one nation allows within its territory to the legislative, executive or judicial acts of another nation, having due regard both to international duty and convenience, and to the rights of its own citizens or of other persons who are under the protection of its laws.

• 278See US Const Art VI, § 2. Federalism is discussed in Part IV.C.
• 279552 US 264 (2008).
• 280Id at 275.
• 281See id at 300 (Roberts dissenting).
• 282Id at 279–80.
• 283See Part IV.B.4.
• 284Current habeas jurisprudence, following the Batorian archetype, stresses finality far more than these other values. See text accompanying notes 237–40.
• 285See Michael Williams, 529 US at 436–37 (“[S]tate judiciaries have the duty and competence to vindicate rights secured by the Constitution in state criminal proceedings.”).
• 286Id at 437, quoting O’Sullivan v Boerckel, 526 US 838, 844 (1999).
• 287See 28 USC § 2254(b)–(c).
• 288Pinholster, 563 US at 182.
• 289Id at 186 (quotation marks omitted), quoting Sykes, 433 US at 90.
• 290Pinholster, 563 US at 181.
• 291See Dennis v Secretary, Pennsylvania Department of Corrections, 834 F3d 263, 289 (3d Cir 2016) (en banc) (refusing to accept “an argument that was not even mentioned by the [state court], much less fairly presented before it”).
• 292See id at 281 (“Richter and its progeny do not support unchecked speculation by federal habeas courts in furtherance of AEDPA’s goals.”).
• 293See William J. Brennan Jr, Federal Habeas Corpus and State Prisoners: An Exercise in Federalism, 7 Utah L Rev 423, 442 (1961) (“The state judiciaries, responsible equally with the federal courts to secure [federal] rights, should be encouraged to vindicate them. A self-fashioned abdication by the federal courts of their habeas corpus jurisdiction . . . would not provide that encouragement.”).
• 294Wilson, 834 F3d at 1265 (Pryor dissenting).
• 295Id (Pryor dissenting), citing Richter, 562 US at 96. California voters recently approved Proposition 66, a portion of which requires capital petitioners to file their initial petitions in the trial court. Cal Penal Code § 1509. These petitioners are guaranteed “a statement of decision explaining the factual and legal basis for its decision.” Cal Penal Code § 1509(f). Petitioners can then appeal adverse decisions, subject to certain limitations. Cal Penal Code § 1509.1. The California Supreme Court upheld the constitutionality of Proposition 66 in Briggs v Brown, 2017 WL 3624094 (Cal).
• 296Wilson, 834 F3d at 1265 (Pryor dissenting).
• 297Id at 1248 (Pryor dissenting).
• 298See id at 1238. See also Mortara Brief at *7 (cited in note 134) (“[T]o apply § 2254(d)(1) to a second-to-last adjudication on the merits would result in an advisory opinion regarding a non-operative decision that has since been supplanted.”). However, under the look-through presumption, a federal court still applies § 2254(d) to the last adjudication on the merits. Arguments about “advisory opinions” conflate applying § 2254(d) directly to the lower court’s written opinion with a presumption that the summary disposition adopts the reasoning below.
• 299See Wilson, 834 F3d at 1239 (“In this way, federal courts can use previous opinions as evidence that the relevant state court decision under review is reasonable.”).
• 300Another aspect deserves brief mention. To grant relief, federal judges are in the unenviable position of calling state decisions not just incorrect, but in fact unreasonable. See Washington v Schriver, 255 F3d 45, 62 (2d Cir 2001) (Calabresi concurring) (describing the “highly undesirable [situation] of having federal courts reviewing State court decisions on habeas frequently declare such decisions to be not just mistaken but also unreasonable”). And, after Richter, federal judges may even have to say that each state judge who denied relief was not “fairminded.” See Reinhardt, 113 Mich L Rev at 1229 (cited in note 14) (explaining that, if Richter’s language were taken literally, “in order to grant habeas relief, [federal judges] would need to find that each of the state court judges who denied the petitioner’s claim was not fairminded”). The use of such critical language does not foster judicial comity, but it would be incomprehensible to deny petitioners relief because the Supreme Court’s chosen verbal formulation of § 2254(d) is emotionally charged.
• 301Donald, 135 S Ct at 1376, quoting Woodford v Visciotti, 537 US 19, 24 (2002) (per curiam).
• 302See, for example, Panetti, 551 US at 953; Early v Packer, 537 US 3, 8 (2002) (per curiam) (stating that § 2254(d) does not preclude relief if either “the reasoning [or] the result of the state-court decision contradicts [Court precedent]”).
• 303Richter, 562 US at 100.
• 304See Dennis, 834 F3d at 353 (Jordan concurring) (“We would do real damage to [comity and federalism] were we to begin re-writing state court opinions to save them.”); Woolley v Rednour, 702 F3d 411, 422 (7th Cir 2012) (“It would be perverse, to say the least, if AEDPA deference required this court to disregard a state court’s expressed rationale for a decision.”).
• 305Bator, 76 Harv L Rev at 451 (cited in note 237).
• 306See Wilson, 834 F3d at 1245 (Jordan dissenting):

  Starting with a result (the result reached in a summary denial of relief), then coming up with hypothetical reasons to support that result, and then assessing whether such imagined reasons are contrary to or an unreasonable application of clearly established Supreme Court precedent, is not what appellate courts normally do. The notion of a court starting with a result, and then searching far and wide for reasons to justify that result, turns the notion of neutral decision­making on its head.

• 307See text accompanying notes 289–93.
• 308Sykes, 433 US at 89.
• 309See Wilson, 834 F3d at 1236:

  It makes sense to assume that a summary affirmance rests on the same general ground—that is, a procedural ground or on the merits—as the judgment under review. . . . But it does not follow that a summary affirmance rests on the same specific reasons provided by the lower court.

  See also Mortara Brief at *11 (cited in note 134) (“And, notably, Ylst never endorses Wilson’s proposed microscopy of assuming the later unexplained order adopted precisely the same reasons . . . , instead looking to the macroscopic ‘grounds’—procedural or not.”).

• 310Ylst, 501 US at 806.
• 311Id.
• 312See id at 804; Cannedy II, 733 F3d at 800 (“Taking into account the circumstances surrounding the state court’s unexplained decisions, the Supreme Court tells us to adopt the most logical explanation for the state court’s actions.”).
• 313Ylst, 501 US at 804.
• 314See id.
• 315See Wilson, 834 F3d at 1236–37 (“[Ylst] does not direct a federal court to treat the reasoning of a decision on the merits by a lower court as the reasoning adopted by a later summary decision that affirms on appeal, especially since neither the Supreme Court nor any federal circuit court operates that way.”).
• 316Id at 1267 (Pryor dissenting).
• 317See id at 1267 n 22 (Pryor dissenting).
• 318See id at 1236–37 (collecting US Supreme Court and Eleventh Circuit opinions that discuss summary affirmances).
• 319See Anderson v Celebrezze, 460 US 780, 784 n 5 (1983) (quotation marks omitted):

  We have often recognized that the precedential effect of a summary affirmance extends no further than the precise issues presented and necessarily decided by those actions. A summary disposition affirms only the judgment of the court below, and no more may be read into our action than was essential to sustain that judgment.

  See also Wisconsin Department of Revenue v William Wrigley, Jr, Co, 505 US 214, 224 n 2 (1992).

• 320If a summary affirmance by the Supreme Court bestowed precedential effect on the opinion from the court below, the potentially incorrect reasoning of a circuit court would become the law of the land without the benefit of a full round of briefing on the merits before the Court.
• 321Cannedy II, 733 F3d at 800–01.
• 322See Wilson, 834 F3d at 1248 (Pryor dissenting).
• 323See id at 1239. See also Seligman, Note, 64 Stan L Rev at 497–98 (cited in note 52) (relaying the remarks of Justice Carlos Moreno of the Supreme Court of California, who described the “only substantial difference” between summary dispositions and written opinions as “the process of converting the in-chambers analysis into a written opinion fit for publication”).
• 324See Wilson, 834 F3d at 1262 (Pryor dissenting).
• 325See Sallie v Sellers, No S17W0685, slip op at 1 (Ga Dec 6, 2016), archived at http://perma.cc/KD7H-UG93 (“Although this Court does not necessarily endorse all of the habeas court’s reasoning, it is clear that the habeas court properly denied relief.”).
• 326See Mortara Brief at *12 n 6 (cited in note 134).
• 327See Ylst, 501 US at 803.
• 328See Tara Williams, 568 US at 300 (“[F]ederal courts have no authority to impose mandatory opinion-writing standards on state courts.”).
• 329Wilson, 834 F3d at 1238.
• 330See id; Richter, 562 US at 99.
• 331See Richter, 562 US at 99.
• 332See Part IV.B.3.
• 333See Huq, 81 U Chi L Rev at 570–81 (cited in note 14).
• 334Id at 570–77.
• 335See id at 571.
• 336See Schriver, 255 F3d at 62 (Calabresi concurring):

  [I]f AEDPA deference were deemed automatically and universally to apply, then that law would require extremely busy State court judges to figure out what can be very complicated questions of federal law at the pain of having a defendant incorrectly stay in prison should the State court decision of these complex questions turn out to be mistaken (but not unreasonably so).

• 337Id at 63 (Calabresi concurring).
• 338Id (Calabresi concurring).
• 339Huq, 81 U Chi L Rev at 577 (cited in note 14). “Judges would have to be motivated by a preference for not being contradicted by a federal judge and would have a sufficiently low discount rate so as to be motivated by the specter of federal habeas relief some ways down the road.” Id at 577–78.
• 340Judge Jill Pryor defends a de minimis exception to the Tara Williams opinion-writing rule, as the state court “could simply issue a one-line order denying an [appeal] . . . that indicates agreement with the result the [lower] court reached but not the lower court’s reasons for rejecting the petitioner’s claim.” Wilson, 834 F3d 1263 & n 18 (Pryor dissenting). Not only can state courts summarily affirm the result while explicitly declining to endorse the reasoning with minimal additional effort, they in fact already do this. See note 325 and accompanying text. Because the Supreme Court, on direct review, looks through when reviewing denials of nondiscretionary review, looking through on federal habeas review only marginally increases the incentive to write an opinion. See text accompanying notes 203–08. See also Sears v Upton, 561 US 945, 953–54 (2010) (looking through a denial of nondiscretionary review pursuant to 28 USC § 1257(a)).
• 341See Huq, 81 U Chi L Rev at 578 (cited in note 14) (“The treatment of summary opinions as merits judgments, for example, effectively imposes a tax on reasoned adjudication by state courts.”) (citation omitted).
• 342Richter, 562 US at 99.
• 343See, for example, Mann v Ryan, 774 F3d 1203, 1225 (9th Cir 2014) (Kozinski concurring in part and dissenting in part) (describing “the perverse effect of encouraging state courts to deny relief summarily, to insulate their orders from tinkering by the federal courts”); Dennis, 834 F3d at 374 (Hardiman dissenting) (“[B]ecause it makes AEDPA deference inversely proportional to the amount of information the state court provides, it creates a perverse incentive for state courts to earn the deference of federal courts by saying less.”). See also Seligman, Note, 64 Stan L Rev at 479 & n 61 (cited in note 52) (“State courts might use summary dispositions not in spite of the difficulty they present to a federal habeas court in granting relief, but because of it.”) (sharing former-Judge Michael McConnell’s worries).
• 344See, for example, Seligman, Note, 64 Stan L Rev at 493 (cited in note 52) (opining that it “may strike some as absurd or even tyrannical” that state courts could seek “safe harbor” by not providing reasons).
• 345See Kovarsky, 82 Tulane L Rev at 456 (cited in note 2).
• 346See Coleman, 501 US at 760 (Blackmun dissenting) (“Federal habeas review of state-court judgments, respectfully employed to safeguard federal rights, is no invasion of state sovereignty.”).
• 347Justice William J. Brennan Jr remarked that “[w]e prize our federalism because of the proved contributions of our federal structure towards securing individual liberty,” noting the ability of federal habeas review to direct states to secure prisoners “against invasion of the rights guaranteed them by the basic law of the land.” Brennan, 7 Utah L Rev at 442 (cited in note 293). See also Bond v United States, 546 US 211, 223–24 (2011) (stating that an individual may have standing to challenge “a law [ ] enacted in contravention of constitutional principles of federalism”).
• 348See Coleman, 501 US at 759 (Blackmun dissenting) (“Federalism . . . has no inherent normative value: It does not . . . blindly protect the interests of States from any incursion by the federal courts.”).
• 349Kovarsky, 82 Tulane L Rev at 446 (cited in note 2).
• 350Hoffmann and King, 84 NYU L Rev at 804–05 (cited in note 7).
• 351Justice Felix Frankfurter called the Great Writ “the basic safeguard of freedom” and “one of the decisively differentiating factors between our democracy and totalitarian governments.” Brown, 344 US at 512 (Frankfurter concurring).
• 352501 US 722 (1991).
• 353Id at 726.
• 354See Huq, 81 U Chi L Rev at 555–56 (cited in note 14).
• 355See id at 556 (“Even Roberts Court jurisprudence evinces some concern for ‘the historic importance of federal habeas corpus proceedings as a method for preventing individuals from being held in custody in violation of federal law’ as a counterweight on the other side of the scales.”), quoting Trevino v Thaler, 133 S Ct 1911, 1916–17 (2013).
• 356133 S Ct 1911 (2013).
• 357See Garrett and Kovarsky, Federal Habeas Corpus at 356 (cited in note 35) (recognizing a “[r]enewed [e]mphasis” on state postconviction review).
• 358136 S Ct 718 (2016).
• 359Id at 731–32.
• 360See Reinhardt, 113 Mich L Rev at 1232 (cited in note 14):

  [T]he fact that resource-constrained state courts have a backlog of cases is not a reason in favor of deference; it clearly cuts in the opposite direction, as truly meritorious claims are far more likely to be missed under a system in which state court judges simply are not able to exercise the same degree of care as federal appellate judges.

• 361See Huq, 81 U Chi L Rev at 557 (cited in note 14) (“Attending solely to one side of the scale yields only incomplete insight because it does not speak to how the scale is calibrated.”).