The Habeas Optimist

Lee Kovarsky

81 U Chi L Rev Dialogue 108 [Essay PDF]
A response to Aziz Z. Huq, Habeas and the Roberts Court, 81 U Chi L Rev 519 (2014), available here.


For those who believe that legal rules are supposed to predictably map events onto outcomes, federal postconviction law is a frustrating mess. Most of those who try to make sense of it end up with some variant of a pretty cynical model: if the claimant is an inmate convicted in state court, then federal relief is unavailable. Some of us, however, remain cautiously receptive to theories that high-court habeas outcomes express a more complex function.1 In Habeas and the Roberts Court, Professor Aziz Huq establishes himself as the field’s foremost academic optimist.2

In Huq’s view, what might seem to be madness on the Roberts Court conceals method. The demand for federal habeas resources outstrips supply, and Huq believes that the Roberts Court has responded by developing two “tracks” of federal postconviction inquiry: (1) a resource-light Track One marked by numerous procedural barriers and an impossible-to-satisfy standard of review; and (2) a resource-intensive Track Two, in which convicted state inmates obtain relatively unimpaired federal habeas review of their constitutional claims. According to Huq, the skeleton key to this jurisprudence is something that I will call the “fault delta.” The fault delta is the difference in blameworthiness between the inmate and the state. Briefly stated, Huq’s descriptive thesis is that the fault delta determines the track along which a habeas petition travels and how it is treated as it moves along the selected path.3 His normative thesis, to which I give very short shrift in the interest of available space, is that the two-track model is more workable and desirable (relatively speaking) than some of the legislative innovations that recent scholarship has promoted.4

In Part I of this Essay, I focus on the defects in Huq’s explanatory account. First, I cannot discern two tracks of federal postconviction process, if a track is defined as a distinct path along which an inmate avoids having to satisfy all potential restrictions on relief. Using two tracks as a descriptive framework, then, does more to confuse than to illuminate the role of fault in the federal postconviction process. Second, even if one accepts that a series of procedural inquiries constitutes a second track, Huq overstates the role of fault in determining how each track operates. Although the concept of state fault explains much about whether a claim subject to a state merits disposition clears 28 USC § 2254(d) (the state-federal relitigation bar), it explains much less about the operation of other restrictions on the habeas remedy.

In Part II, I sound my own optimistic note. (Sort of.) I agree that the habeas jurisprudence of the Roberts Court has certain animating principles; I just think that they differ from those identified by Huq. Instead of a decision tree branched by reference to fault, the Roberts Court has constructed what I call an “on-the-merits” paradigm of federal habeas process: a sequential set of inquiries designed to ensure that a diligent state inmate receives at least one merits disposition on a constitutional claim. Under the on-the-merits paradigm, whether the merits disposition comes in state or federal court does not matter. The on-the-merits paradigm represents a thin procedural commitment. On the one hand, the Roberts Court is increasingly committed to the ideal that diligent prisoners should not lose claims without some merits disposition. On the other hand, it appears to have deserted the principle that a state merits disposition should be the product of reliable process.

I.  The Role of Fault in a Single Track

According to Professor Huq, the Supreme Court has constructed two tracks of federal habeas law, with the scrutiny afforded in each track determined by reference to the fault delta.5 Huq has, by my lights, somewhat overstated fault’s role as an organizing principle. First, habeas process is not bifurcated into multiple tracks. The multitrack model erroneously assumes that a track terminates with a different inquiry, each corresponding to a different standard for scrutinizing the merits of a federal claim. In reality, there are not two such “standards of review.” (I will explain the scare quotes in a moment.)

Second, although the concept of fault does significant explanatory work, it does so only in cases in which a state decision is “on the merits.” Specifically, if a state court is “at fault” whenever it badly fumbles merits processing of the legal and factual data before it, then fault is the lodestar for one of the most important remedial restrictions that federal courts adjudicate. But not all state decisions are merits dispositions; when there is federal habeas litigation over state nonmerits dispositions, the role of fault is diminished.

A.    A Single Track

My first major objection to Huq’s theory that there is a two-track habeas regime organized around fault is that the Roberts Court has not in fact constructed two tracks. If a path is a line connecting discrete points, and if each point is a condition for federal habeas relief, then separate tracks would require either multiple end points or multiple routes from the first point to the last one. A map of Roberts Court decision logic, however, discloses only a single path. For that reason, a two-track model actually obscures the active role of fault in modern habeas law.

1.     Defining multiple tracks.

Huq would identify each of the two federal habeas tracks with a different “standard of review,” applicable at two different terminal points on a logic map. (If the term “logic map” does not make intuitive sense, its meaning should be evident in Figures 1 through 3 below.) For Huq, Track One defines habeas process for “most petitions that are either adjudicated on the merits in state court or, instead, subject to adequate and independent state bars or, alternatively, federal procedural constraints.”6 Track Two, by contrast, defines habeas process for inmates satisfying cause-and-prejudice inquiries necessary to excuse an otherwise-disqualifying procedural defect in the claim.7 A federal court concludes the Track One process by applying a deferential “standard of review” to the order denying relief on the constitutional claim, and it concludes the Track Two process with de novo “review.”8

Perhaps the easiest way to convey the operation of the Roberts Court’s habeas rules is through a visual representation. Figure 1 is a logic map that depicts a simplified, single track of federal habeas process. I believe that the logic map in Figure 1 captures the doctrine more accurately than the logic maps depicted in Figures 2 and 3.

Figure 1.  Single-Track Federal Habeas Process


If there is more than one track, however, then there should be multiple paths from the starting point to one or more end points. First, there may be multiple tracks because there is more than one end point. Figure 2 depicts a logic map with multiple terminal points, and this multitrack model is the one that I consider most carefully here.

Figure 2.  Multitrack Federal Habeas Process—Multiple End Points


Second, there may be multiple tracks because there is more than one path to the same end point. Figure 3 depicts a multitrack map with a single terminus.

Figure 3.  Multitrack Federal Habeas Process—Single End Point


Huq favors syntax over pictures but seems to argue that the Roberts Court is developing decision logic along the lines of that depicted in Figure 2—that is, a multitrack jurisprudence with different terminal inquiries. Specifically, Huq’s multiple references to different “standards of review” at the terminal points of each track lead me to believe that he is not theorizing a multitrack, single-end-point logic map like that in Figure 3.9 Without getting bogged down in detail, even if Figure 3 did capture Huq’s theory, it would be vulnerable to the same criticisms that I level at the multiple-end-points model.10

2.     Section 2254(d) and preclusion.

Federal habeas process does not have two different tracks resulting in two different terminal inquires, because there is in fact only one terminal inquiry. Phrased visually, the Roberts Court’s doctrine is captured most accurately by Figure 1, not Figure 2. What Huq describes as the “standard of review” under § 2254(d) is actually a relitigation restriction that any successful inmate must avoid, and any inmate avoiding it must still satisfy the single terminal inquiry: the inmate must prove the merits of the constitutional claim.

Section 2254(d) restricts federal habeas relief for claims lodged by state inmates.11 Courts, lawyers, and academics almost always refer to § 2254(d) as a (substantive) “standard of review,”12 but § 2254(d) behaves much more like a (procedural) preclusion rule. (The scare quotes should make sense now.) If § 2254(d) were a standard of review, a state inmate would get relief by satisfying it; but that is not how the process works. If § 2254(d) indeed behaves more like a preclusion rule, then there are not two terminal points on the logic map. There is a relitigation bar that all successful inmates must avoid, and the terminal inquiry for all inmates is the straightforward merits question. Fault may matter a great deal, but the Roberts Court has not organized two tracks of habeas review around it.

The full argument for treating § 2254(d) as a procedural relitigation restriction is too long for this space, but its basic contours are accessible enough. Section 2254 is the major section of the US Code dealing with federal habeas process for state inmates. Subsection 2254(a) is the basic grant of federal habeas power to relieve unlawful state custody. Other § 2254 subsections restrict the availability of the federal remedy. For example, § 2254(b) disables the federal habeas remedy in the event that an inmate has not exhausted state process. Section 2254(d) is the relitigation restriction on the habeas remedy—that is, a statutory preclusion rule based on the state decision. It is not a standard for reviewing the merits of the underlying claim. As I explain in Part I.4, the Roberts Court is increasingly prone to this construction of the statute.

3.     The two-track theory and the relitigation bar.

If § 2254(d) is treated as a (procedural) relitigation restriction rather than a (substantive) standard of review, then the second track vanishes. The starting point is filing and the end point is merits review, and the inmate must satisfy every procedural restriction in between. Because an inmate never bypasses a restriction by satisfying another one, there is simply one track, with a lot of stops. If a claim is excused from procedural default, then the claimant still has to avoid the limitations bar, the successive-petition restriction, and § 2254(d); if a claim is not time-barred, then the claimant must still avoid procedural default, the successive-petition restriction, § 2254(d), and so forth.

Of course, even if outcomes do not form a separate track, they can be highly correlated. For example, inmates excusing the procedural default of a claim usually do avoid the § 2254(d) relitigation bar, but not because of the fault delta. Instead, the outcomes correlate (and inmates obtain federal merits review) because § 2254(d) restricts relief only for claims that a state court decides “on the merits.”13 A claim is procedurally defaulted only when the state ground for rejecting the claim was “adequate and independent.”14 The independence requirement, in turn, almost always ensures that the state ground underlying a default is a nonmerits disposition. Inmates who can excuse the default can obtain merits review, but the showing necessary to secure the excuse has nothing to do with how the claim is processed under § 2254(d). Inmates with procedurally defaulted claims can avoid § 2254(d) because the ground for denying state relief was, by definition, not on the merits. Otherwise, the claim would not be procedurally defaulted to begin with.

If the absence of inmate fault is driving the § 2254(d) outcomes—more so than the presence of a nonmerits adjudication—then one would expect inmates found to be diligent pursuant to other procedural inquiries to obtain favorable § 2254(d) outcomes. The evidence is not there. This short Essay does not lend itself to comprehensive treatment of a controlled test, but one example should suffice. The statute of limitations in § 2254(d) is equitably tolled by a showing of extraordinary circumstances (cause, roughly) and inmate diligence.15 If the fault showing were the variable entitling an inmate to relief on the merits, then one would expect a finding of equitable tolling to entail consideration on the merits. Unfortunately, a finding of equitable tolling—even if the “extraordinary circumstances” that trigger it are purely state-fault conditions—does not allow an inmate to avoid the § 2254(d) relitigation bar.

4.     Evidence from the Supreme Court.

The Roberts Court has played an active role in preserving the most important condition for a single-track model—that § 2254(d) behaves more like a relitigation restriction than a standard of review. The Roberts Court’s most obvious fingerprint is Cullen v Pinholster,16 which brought lower-court practice into line with the view that § 2254(d) is a relitigation restriction and not a standard of review.17 Before Pinholster, many federal courts—consistent with a very credible reading of § 2254(d)(1)—evaluated the reasonableness of state merits dispositions by reference to evidence outside the state record.18 (In other words, those courts analyzed the merit of the claim.) Pinholster changed that, bringing § 2254(d) into line with common-law relitigation rules in which “the record” of the initial decision precluded collateral litigation.19 Pinholster is more broadly representative of the relitigation framework that the Roberts Court uses to construct § 2254(d).20 Any traces of multitrack jurisprudence remain in spite of, not because of, the Roberts Court.

B.    The Narrower Significance of Fault

Although Huq overstates the centrality of fault to modern habeas jurisprudence, I agree with his assessment of several nonfault variables. Specifically, I share Huq’s skepticism about federalism and feedback loops.21 “Federalism” is more of a decisional idiom that correlates with habeas outcomes than an actual source of outcome causation. State criminal-justice administration is tragically underresourced and error prone, but that state of affairs is not particularly sensitive to recent or contemplated changes in federal postconviction law.

I have two major disagreements with the fault-based account. First, the analogy to constitutional tort works only for the § 2254(d) inquiry. With respect to the other procedural inquires, fault means something materially different. Characterizing these inquires as fault-based requires an elastic definition of fault that undermines Huq’s explanatory objective. Second, Huq has excluded several areas of habeas doctrine that rather decisively demonstrate a commitment to sorting cases based on innocence—a sorting principle he dismisses based on the data that he did not exclude.

1.     Fault in the remedial restrictions.

Because I do not believe that there are discrete tracks of habeas process, I relate my thoughts by reference to the specific procedural inquiries marking any potential path to federal relief. Huq argues that “postconviction jurisprudence has moved into alignment” with constitutional tort law, by which he means the qualified immunity doctrine that imposes a heightened culpability requirement on any government official sued for damages.22 Habeas doctrine, Huq suggests, is beginning to exhibit many features of other remedial jurisprudence.23 That Huq is mining remedies for transsubstantive principles is itself a noteworthy academic development, but his pursuit of big-picture fit suppresses some important differentiation as expressed through the details.24

The argument that habeas jurisprudence exhibits remedial transsubstantivity is most persuasive if limited to a claim about how the § 2254(d) relitigation restriction works. Federal merits review is available only if an inmate shows that the state court is “at fault” because it unreasonably processed the data before it, with “data” consisting of both law and facts. A state court is at fault when it unreasonably bumbles a decision just like a law-enforcement officer is liable when he or she violates clearly established law.25

If fault is defined that way, then there is indeed some transsubstantive convergence around the fault delta that Huq specifies. First, the Roberts Court has substantially increased the showing of error that an inmate must make in order to qualify a state decision as unreasonable under § 2254(d). Roberts Court opinions regularly imply that a state outcome is not unreasonable unless every “fairminded jurist” would reject it,26 but the Rehnquist Court expressly rejected a functionally equivalent standard when presented with the question in Terry Williams v Taylor.27 Second, in Pinholster, the Roberts Court limited the assessment of legal reasonability to the record before the state court.28 By limiting the § 2254(d) inquiry to the state court record, the Roberts Court opted for a fault-identifying rule rather than an error-identifying one.

Huq’s model has some trouble when it moves beyond § 2254(d) to other restrictions on the habeas remedy. Fault means something a little different for the other restrictions, and those distinctions have two important implications. First, the other definitions of fault undermine a theory of transsubstantive convergence. Second, by using fault as a catchall term for all forms of error, culpability, and diligence that determine habeas outcomes, Huq defines it so abstractly that it ceases to effectively map events onto outcomes.

Specifically, Huq uses fault in at least four senses: (1) the wrongfulness of the underlying constitutional violation during the state criminal proceeding, (2) state culpability exhibited while litigating the violation, (3) state culpability as expressed through a manifest judicial error in adjudicating the claim, and (4) a lapse in inmate diligence in litigating the claim.29 Constitutional-tort law and habeas doctrine converge somewhat with respect to the fault embedded in the § 2254(d) rules—the third sense enumerated above—but not with respect to the other fault categories. Moreover, the breadth of the term undermines the degree to which it is capable of contributing to a precise model. To phrase the point a little more formally, an independent variable cannot usefully predict outcomes when that variable is defined in a way that makes assigning a value to it difficult.

2.     Sorting for innocence.

Huq also excludes from his data many decisions that undermine the proposition that the emerging habeas regime is insensitive to innocence. Specifically, he excludes evidence of a developing, transdoctrinal innocence focus appearing in procedural-default, statute-of-limitations, and successive-petition inquiries. If an assessment of the Roberts Court includes these developments, then the picture changes substantially.

First, Huq excludes miscarriage-of-justice cases from the procedural-default jurisprudence. A miscarriage-of-justice showing—usually threshold proof of actual innocence—excuses the forfeiture in state proceedings.30 He omits those cases because the excuse has a low success rate,31 but all habeas arguments have a low success rate. In any event, the exclusion of the cases seems odd because Huq’s professed point of emphasis is what the Supreme Court is deciding rather than what lower courts are doing.32 Whatever the ground-level success rate, the Roberts Court has decisively (unanimously) preserved the miscarriage-of-justice excuse in the face of arguments that the Antiterrorism and Effective Death Penalty Act33 (AEDPA) abolished it.34 As Huq acknowledges in a footnote, the miscarriage-of-justice excuse plays a pivotal role in facilitating merits review in DNA exoneration cases.35

Second, Huq excludes § 2244(d) limitations cases because they “represent[ ] less an emanation of some deeply felt judicial principle than the Court’s necessary scrimmaging with a poorly drafted rule encountering a heterogeneous set of external circumstances.”36 That observation may be true, but it should not be disqualifying. That courts must construct legal meaning out of encounters between a poorly crafted statute and varied fact patterns is a norm of habeas law, not an exception. For example, Huq’s above-quoted language describes § 2254(d) just as well as it describes the limitations provision. The excluded Roberts-era cases are important because they express an unmistakable commitment to innocence-related exceptions. After creating an equitable-tolling exception to excuse otherwise untimely claims,37 the Court clarified that sufficiently strong evidence of innocence would trigger that unwritten safety valve.38

Third, Huq fails to consider the innocence exceptions to restrictions on successive petitions. In In re Davis,39 the Supreme Court reviewed an order denying authorization for successive habeas proceedings notwithstanding a statutory provision that expressly forbids certiorari consideration.40 The Supreme Court reviewed the actual innocence claim pursuant to its obscure power to issue original habeas writs and found the innocence evidence sufficient to justify an order transferring the case to a district court for factfinding.41 Moreover, that order short-circuited other habeas law in fairly breathtaking ways. It did not require that the district court apply § 2254(d), nor that it find any constitutional violation at all. The transfer simply directed the district court to evaluate innocence.

In isolation, each case thread might be subject to disqualifying caveats. Taken together, however, these cases plainly express a commitment to innocence as an organizing doctrinal principle on the Roberts Court. Fault is part of the story, but any model that marginalizes innocence as a sorting priority is incomplete.

II.  An Alternate Model

In the balance of this response, I provide an alternative account of the Roberts Court’s habeas jurisprudence. Like Professor Huq, I believe that the jurisprudence expresses a discernable paradigm of federal habeas process; the two of us differ only on the particulars. The Court is constructing what I call a “merits-opportunity” regime, which ensures that a diligent state inmate gets at least one shot at a merits adjudication in a jurisdictionally competent court. The merits-opportunity regime differs from earlier models of habeas reform in that it focuses only on the availability of merits disposition, and not on whether the state afforded process that tends to produce accurate outcomes.

Virtually every rule and proposed reform for federal postconviction review bears the imprint of a 1963 Harvard Law Review article by Professor Paul M. Bator, Finality in Criminal Law and Federal Habeas Corpus for State Prisoners.42 Bator proceeded from the epistemic proposition that absolute legal truth was unknowable. He reasoned that federal courts should treat decisions as final when state “processes previously employed for determination of questions of fact and law were fairly and rationally adapted to that task.”43 He believed that institutions ensure accuracy by proxy of reliable procedure and urged that a state merits disposition precludes subsequent federal review as long as the inmate received “full and fair” state process on the claim.44 Bator thought that the full-and-fair model accurately described the history of state-inmate process and that it produced desirable outcomes.45

Rules of Batorian provenance emphasize a finality interest triggered by reliable state-claim processing. The archetypal federal claim in Bator’s day, however, bears no resemblance to the mine-run claim subject to modern habeas process. That change has prompted federal judges and legislators—once largely committed to a round of federal habeas process for all diligently asserted claims—to pursue objectives even more modest than those urged by Bator. Modern habeas paradigms must account for the gradual-but-unmistakable convergence of two phenomena: (1) an exploding set of potential claims, and (2) more complicated claim-forfeiture scenarios. The Roberts Court has responded with a principle that diligent prisoners should get a merits disposition, but it has not committed to the full-and-fair process principles that are hallmarks of Batorian models.

A.    The Decline of the Batorian Archetype

Bator’s model embodies the principle that a diligent inmate should have at least one full-and-fair opportunity to litigate a constitutional challenge before a jurisdictionally competent court.46 Whether the court was state or federal did not matter. The idea was that, in a world in which absolute truth is unknowable, reliable process produces outcomes that ought not to be relitigated in a federal habeas forum. When state courts had supplied such process, Bator believed that federal habeas review added no meaningful increment of confidence in an outcome.

The full-and-fair model was too restrictive to ever secure a legislative majority—even as part of the Republican-crafted AEDPA, the habeas law passed in the immediate wake of the Oklahoma City bombing.47 Notwithstanding legislative setbacks, however, those favoring severe federal habeas restrictions have scored a series of decisive post-AEDPA Supreme Court victories. Not only has the Court endorsed a maximally broad view of what qualifies as a state merits disposition, but it also permits state merits dispositions to preclude federal relitigation notwithstanding many questions about whether underlying state procedure was full or fair.48

A changed paradigm of habeas process was almost inevitable given the changed attributes of the typical state-inmate litigation. Bator’s was the dominant model of federal habeas process because of how elegantly it prescribed outcomes for the mine-run litigant. Bator’s archetype was a state inmate who used federal habeas relief as a vehicle to lodge a constitutional challenge that was litigated at trial, such as a coerced-confession claim.49 Such serial consideration invited the question whether a federal forum should afford additional process to decide a claim that the state might have decided quite reliably.

This inmate archetype—an effectively represented one, capable of lodging the claim at trial—is modern fiction wearing a jumpsuit and bracelets. At least two important things have changed since Bator launched his broadside against the Warren Court’s habeas jurisprudence. First, the number of cognizable constitutional claims has exploded, and many of them are incapable of trial adjudication. Second, Bator wrote at a time when state postconviction (collateral) review was in its infancy, so the conditions of nonmerits adjudication were quite different. Both developments have put enormous institutional pressure on Congress and the Supreme Court to figure out new ways to husband limited federal habeas resources.

Bator published Finality in Criminal Law in January 1963, before the Supreme Court undertook most of what is now described as the “revolution” in modern criminal procedure.50 There was no Brady v Maryland,51 which now requires the prosecution to disclose exculpatory evidence.52 There were no decisions incorporating Sixth Amendment compulsory process,53 jury trial,54 speedy trial,55 and Confrontation Clause rights.56 The Supreme Court had yet to guarantee legal representation to indigent defendants under Gideon v Wainwright57 and was two decades away from specifying how the right to a lawyer required effective assistance of counsel.58 The volume and the content of claims have changed in ways that date Bator’s archetype. The principle that every claim should get a full-and-fair hearing was an easier legal commitment before the criminal-procedure revolution than it is now. Moreover, the coerced-confession heuristic is misleading because the vast majority of modern claims involve information outside the record presented to a state court for consideration—because, for example, the prosecution suppressed evidence (Brady)59 or the defendant’s lawyer was constitutionally ineffective (Strickland v Washington).60 Whether the discussion is about a habeas paradigm’s descriptive accuracy or its normative justification, it cannot proceed persuasively by assuming aberrant claim content.

Nor should the favored paradigm assume that claim content is usually decided on the merits. The full-and-fair ideal flourished at a time when a state inmate was, in fact, far more likely to receive a state merits disposition after considerable process. Modern state deliberation, however, usually entails only cursory procedure. Because so many claims involve content outside the trial record and must therefore be exhausted collaterally,61 state inmates find themselves in increasing contact with state postconviction process, a toxic swamp of byzantine procedure and under-resourced indigent representation. For claims that states must consider collaterally, there is no federal right—statutory or constitutional—to a lawyer, let alone a competent one.62 Pro se inmates end up forfeiting claims in the teeth of obscure state postconviction rules that confound even experienced lawyers. And even when state collateral process culminates in a merits determination, that determination is frequently based on a record constructed under precisely the same constraints that cause mass forfeiture.63 In much the same way that differences in claim content have forced changes in habeas law, so too have differences in the way that content is processed in state collateral proceedings.

B.    The Merits-Opportunity Regime

In a world of finite federal habeas resources, the on-the-ground reality of criminal-justice administration is grim. Mass incarceration follows from the “war on drugs,” constitutional error infects trials because state governments are as resource-strapped as their federal counterpart, and inmates are generally unrepresented when they seek state remedies.64 Huq and I both read the Roberts Court’s habeas jurisprudence as an attempt to separate wheat from chaff in this decaying ecosystem, but we read it in different ways.

Huq overemphasizes the role of fault in how habeas law processes claims; the primary sorting mechanism is not based around the presence or absence of fault but instead reflects whether there is a state merits decision. As a concluding proposition, I want to show that the Roberts Court is developing an on-the-merits paradigm, which involves: (1) committing to a principle that diligent prisoners should have one merits disposition of a claim, and (2) abandoning the principle that the state process producing that merits disposition be full and fair.

Under Bator-influenced models, full-and-fair state merits adjudication precludes federal relitigation. The Roberts Court treats state merits dispositions differently in two ways, with the first effect dominating the second and with the result being that a state merits disposition enjoys greater preclusive scope than what a full-and-fair model would require.

First, under the Roberts Court’s jurisprudence, a state decision triggers the federal preclusion provision if it is on the merits, notwithstanding any failure of the full-and-fair condition.65 There are situations in which a procedural defect in the state proceeding might allow an inmate to avoid the preclusion bar, but such exceptions are limited to failures of state process that themselves rise to the level of constitutional-due-process violations.66

Second, courtesy of the Roberts Court, the existence of a state merits determination makes federal review almost impossible to obtain. The text of the statute does not lead inexorably to that conclusion, permitting federal merits consideration when the state decision was contrary to clearly established law, unreasonably applied clearly established law, or based on an unreasonable determination of fact.67 The Roberts Court, however, has constructed the § 2254(d) restriction in a way that is impossible for all but the most aggrieved petitioner to avoid. Under Pinholster, a state inmate cannot go beyond the state record to show that any of the criteria are satisfied.68 Moreover, in a series of recent decisions, the Roberts Court has resuscitated a no-fairminded-jurist standard for reasonableness that the Rehnquist Court rejected in Terry Williams.69 And, under Harrington v Richter,70 an unreasoned (summary) state order precludes federal relitigation if any hypothetical rationale might be reasonable within the meaning of § 2254(d)(1).71

The foregoing scenarios involve federal habeas treatment of claims subject to a state merits disposition, but the Roberts Court has also developed new rules for dealing with claims that states denied on nonmerits grounds. Huq provides a more detailed account of the developments,72 but I’ll provide a skeletal discussion here. For years, inadequate state postconviction representation could not excuse a default. More specifically, under Coleman v Thompson,73 state postconviction lawyering could never constitute cause under the cause-and-prejudice showing necessary to establish the excuse.74

Coleman was particularly harsh when the underlying challenge was an ineffective-assistance-of-counsel (“IAC”) claim.75 IAC claims are almost universally lodged on collateral review, when the inmate can develop an appropriate record.76 The problem is that state inmates have no federal right to state postconviction counsel and therefore lack any derivative entitlement to an effective lawyer, so the underlying IAC claims were being defaulted constantly.77 Particularly troubling was that the claims were being defaulted without any lapse in inmate diligence—the inmate had simply sustained the double whammy of having had inadequate trial and state postconviction assistance.

In 2012, the Supreme Court decided Martinez v Ryan,78 holding that inadequate state postconviction representation could in fact excuse an otherwise-forfeited IAC claim.79 The result is that—for the single most frequently asserted constitutional violation—the Roberts Court established a device for diligent inmates without a state merits disposition to obtain federal merits review. IAC claims, however, are not the only claims that must be litigated collaterally. For example, claims that the prosecution withheld exculpatory evidence or knowingly elicited false testimony must be litigated under precisely the same conditions as IAC claims.80 As a result, the fit between my on-the-merits model and the current reality of postconviction litigation is imperfect—although justices have already identified the possibility that the Court might move in a direction that improves it. Dissenting in Martinez, Justice Antonin Scalia (correctly) argued that nothing distinguished the rationale that the Court used to permit merits review of the underlying IAC claim from a rationale that it would use to permit merits review of any other claim that must necessarily be litigated collaterally.81 Despite the imperfections in the model, I nonetheless prefer it to Huq’s because, in my estimation, it generates better predictions with abstractions that suppress less variation.82

Although my model involves fault, it has nothing to do with the fault of the state; it involves the fault of the inmate. A state inmate is permitted to obtain merits review of an otherwise forfeited claim if there was no lapse in diligence, and the culpability of the state does not matter. Indeed, even the doctrinal positioning of Martinez—permitting inadequate state postconviction representation to constitute cause83—emphasizes that the excuse is based on a forfeiture caused by something external to the inmate. If state fault really had anything to do with it, then the first substantive claim subject to a modified Coleman rule would not be an IAC challenge, but rather a Brady violation. Whereas IAC challenges involve deficient performance by court-appointed lawyers, Brady violations involve active suppression by the State.

* * *

Huq plots modern habeas jurisprudence along two paths, with fault sorting inmates both between and within each track. Modern jurisprudence is better conceptualized as a single track, with each stop on that track representing a different remedial restriction. As expressed through several of these restrictions, one unmistakable principle emerging from the decisional law is the importance of ensuring that a diligent inmate have an opportunity to at least one on-the-merits adjudication, whether it be in state or federal court.


I describe Professor Huq as “The Habeas Optimist” for two distinct reasons. First, as a descriptive matter, Huq sees in the Roberts Court’s habeas jurisprudence decisional patterns that meaningfully map facts onto outcomes. Second, as a normative matter, he sounds a not-completely-dispirited note, suggesting that habeas activity on the Roberts Court might be desirable insofar as it brings attention to the pathetic condition of indigent representation, amplifies death penalty discourse, and catalyzes political responses through the very act of review.84 Although I share Huq’s skeptical view of legislative-reform proposals, I still fall into the category of people for whom his normative position is “the squeezing of sour lemonade from withered lemons.”85 I comment on that position only briefly and in conclusion.

Whatever the symbolic value of the Supreme Court’s contemporary habeas decisions, the regime will never make sense ethically or economically until the Court acknowledges and addresses certain basic realities. Whether an inmate obtains merits review is hugely sensitive to differences in postconviction representation and barely sensitive at all to the quality of the underlying claim. The habeas jurisprudence soft pedals some of the most obvious pathologies of state law enforcement and criminal-justice administration, favoring instead a wishful parity principle that state and federal governments tend to administer federal law with equal zeal. Whether the Roberts Court is developing habeas jurisprudence in one track or two, the doctrine is still a train wreck.      

          Associate Professor, University of Maryland Francis King Carey School of Law. Thank you to Aziz Huq and to participants at the University of Maryland Carey Law Junior Faculty workshop for invaluable comments on various drafts of this Essay.

      1     As a formal matter, a habeas proceeding is a challenge, heard in court, to the lawfulness of custody. The form of custody at issue in a habeas proceeding need not be a state conviction, or even a criminal conviction at all. Unless I indicate otherwise, however, I intend the terms “habeas” and “postconviction” to operate interchangeably.

      2     See generally Aziz Z. Huq, Habeas and the Roberts Court, 81 U Chi L Rev 519 (2014).

      3     See id at 528.

      4     See id at 594.

      5     See id at 525–28.

      6     Huq, 81 U Chi L Rev at 528–29 (cited in note 2).

      7     See id at 529.

      8     See id at 535–36 (assigning “highly deferential standards of review” for Track One processing); id at 549 (assigning “plenary review” for Track Two processing).

      9     See, for example, id at 535.

      10   Even if Huq believes § 2254(d) to be a relitigation bar rather than a standard of review, then an inmate avoiding it would still need to satisfy every other restriction in the jurisprudence.

      11   28 USC § 2254(d).

      12   See, for example, 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, 697 (2003).

      13   28 USC § 2254(d).

      14   Huq, 81 U Chi L Rev at 543 n 111 (cited in note 2) (“Default rests on an adequate and independent state ground. The independence prong ensures that a claim found to be defaulted is never ‘on the merits.’”).

      15   See id at 544 n 104 (“The habeas statute of limitations is a creature of Congress. 28 USC § 2244(d)(1). The Court, however, has supplemented that statute with an equitable-tolling exception.”); Holland v Florida, 560 US 631, 649 (2010) (concluding that a “[habeas] petitioner is entitled to equitable tolling only if he shows (1) that he has been pursuing his rights diligently, and (2) that some extraordinary circumstance stood in his way and prevented timely filing”) (quotation marks omitted).

      16   131 S Ct 1388 (2011).

      17   See id at 1398 (holding that “review under § 2254(d)[ ] is limited to the record that was before the state court that adjudicated the claim on the merits”).

      18   See, for example, Wilson v Mazzuca, 570 F3d 490 (2d Cir 2009); Pecoraro v Walls, 286 F3d 439 (7th Cir 2002); Valdez v Cockrell, 274 F3d 941 (5th Cir 2001).

      19   Pinholster, 131 S Ct at 1400.

      20   See, for example, Greene v Fisher, 132 S Ct 38, 44 (2011) (describing § 2254(d) as a “relitigation bar”); Harrington v Richter, 131 S Ct 770, 785 (2011) (same); Premo v Moore, 131 S Ct 733, 739 (2011) (specifying when § 2254(d) “permit[s] relitigation”).

      21   See Huq, 81 U Chi L Rev at 558, 581 (cited in note 2) (asserting that “theories based on feedback loops provide no compelling normative warrant for the doctrinal status quo”).

      22   Id at 581.

      23   See id at 583–85.

      24   Of course, virtually no positive social science model built around abstracted predictors can be useful “in the real world” without suppressing the detail necessary to make the abstraction. Nothing about such abstraction is fatal to a predictive model. See Milton Friedman, The Methodology of Positive Economics, in Milton Friedman, ed, Essays in Positive Economics 3, 40–41 (Chicago 1962). One predictive model might nonetheless be preferable to another because, for instance, it either has equal predictive power with less detail suppression or has more predictive power and equivalent detail suppression. See id at 10.

      25   There are already problems in equating constitutional violations of state law enforcement with the adjudication of those in a court, but I set those aside in the interest of space.

      26   See, for example, Richter, 131 S Ct at 786.

      27   529 US 362, 376–77 (2000).

      28   See Pinholster, 131 S Ct at 1398.

      29   See Huq, 81 U Chi L Rev at 583–84 (cited in note 2).

      30   See Schlup v Delo, 513 US 298, 314–15 (1995).

      31   See Huq, 81 U Chi L Rev at 525, 533 (cited in note 2)

      32   See id at 524.

      33   Pub L No 104-132, 110 Stat 1214 (1996).

      34   See, for example, House v Bell, 547 US 518, 536 (2006).

      35   See Huq, 81 U Chi L Rev at 533 n 59 (cited in note 2). In fact, the miscarriage-of-justice excuse is crucial to any innocence claim, not just those predicated on DNA evidence.

      36   Id at 525.

      37   See Holland, 560 US at 645.

      38   See McQuiggin v Perkins, 133 S Ct 1924, 1931 (2013).

      39   557 US 952 (2009).

      40   Id at 952.

      41   Id.

      42   See generally Paul M. Bator, Finality in Criminal Law and Federal Habeas Corpus for State Prisoners, 76 Harv L Rev 441 (1963). Many late twentieth-century legislative proposals derived from Bator’s model. See, for example, Lee Kovarsky, AEDPA’s Wrecks: Comity, Finality, and Federalism, 82 Tulane L Rev 443, 460–62, 502−07 (2007). The Supreme Court regularly cites Bator’s idea to support the promotion of a finality interest. See, for example, Calderon v Thompson, 523 US 538, 555 (1998); McCleskey v Zant, 499 US 467, 492 (1991); Teague v Lane, 489 US 288, 309 (1989).

      43   Bator, 76 Harv L Rev at 455 (cited in note 42).

      44   Id at 456.

      45   See id at 451–53, 463–65. Huq correctly questions several basic justifications for Bator’s model, including the notion that plenary review in federal court would “demoralize” state judges. See Huq, 81 U Chi L Rev at 572–73 (cited in note 2).

      46   By “jurisdictionally competent,” I simply mean a court capable of exercising personal jurisdiction over the respondent-custodian and subject-matter jurisdiction over the pertinent federal questions (the claims).

      47   See Benjamin R. Oyre III, The Failure of Words: Habeas Corpus Reform, the Antiterrorism and Effective Death Penalty Act, and When a Judgment of Conviction Becomes Final for the Purposes of 28 U.S.C. § 2255(1), 44 Wm & Mary L Rev 441, 452–53 (2002).

      48   See Parts I.A.4, II.B.

      49   See Bator, 76 Harv L Rev at 526–27 (cited in note 42).

      50   Corinna Barrett Lain, Countermajoritarian Hero or Zero? Rethinking the Warren Court’s Role in the Criminal Procedure Revolution, 152 U Pa L Rev 1361, 1389 (2004).

      51   373 US 83 (1963).

      52   Id at 87, 90–91.

      53   See Washington v Texas, 388 US 14, 19 (1967).

      54   See Duncan v Louisiana, 391 US 145, 149–50 (1968).

      55   See Klopfer v North Carolina, 386 US 213, 225–26 (1967).

      56   See Pointer v Texas, 380 US 400, 407–08 (1965).

      57   372 US 335, 340 (1963).

      58   See Strickland v Washington, 466 US 668, 684–86 (1984).

      59   See Brady, 373 US at 84.

      60   466 US 668 (1984). The leading empirical work on habeas litigation in district courts found that ineffective-assistance-of-counsel claims were raised in 81 percent of capital cases and 50.4 percent of noncapital cases and that Brady (or related) claims were raised in 43.1 percent of capital cases and 13 percent of noncapital cases. Nancy J. King, Fred L. Cheesman II, and Brian J. Ostrom, Final Technical Report: Habeas Litigation in US District Courts: An Empirical Study of Cases Filed by State Prisoners under the Antiterrorism and Effective Death Penalty Act of 1996 *28, 30 (2007), online at (visited Nov 12, 2014).

      61   See Eve Brensike Primus, Effective Trial Counsel after Martinez v Ryan: Focusing on the Adequacy of State Procedures, 122 Yale L J 2604, 2609 (2013).

      62   See Kovarsky, 82 Tulane L Rev at 466 (cited in note 42) (noting that “[t]here is no federal requirement that offenders have effective counsel during any state collateral review”). There are some state laws requiring that some inmate categories be represented. These laws almost always involve capitally sentenced inmates. See, for example, Overview of Capital Punishment Laws, Washington Courts (2014), online at (visited Nov 14, 2014) (noting that, beginning in 1997, “the Washington State Supreme Court began requiring the appointment of defense co-counsel for death penalty cases at trial and on appeal, consistent with American Bar Association and federal law guidelines”). See also Eric M. Freedman, Giarranto is a Scarecrow: The Right to Counsel in State Capital Postconviction Proceedings, 91 Cornell L Rev 1079, 1086 n 45 (2006) (collecting statutes).

      63   See, for example, Martinez v Ryan, 132 S Ct 1309, 1317 (2012) (noting that, “[w]hile confined to prison, the prisoner is in no position to develop the evidentiary basis for a claim of ineffective assistance, which often turns on evidence outside the trial record”).

      64   See Huq, 81 U Chi L Rev at 588–90 (cited in note 2) (discussing the problems associated with mass incarceration in the American criminal-justice system that has persisted since the 1970s).

      65   See id at 536–37.

      66   See, for example, Panetti v Quarterman, 551 US 930, 953 (2007) (holding that an inmate avoided § 2254(d) by showing an “antecedent” due-process violation).

      67   See 28 USC § 2254(d).

      68   See Part I.A.4.

      69   See Huq, 81 U Chi L Rev at 538–39 (cited in note 2).

      70   131 S Ct 770 (2011).

      71   Id at 786.

      72   See Huq, 81 U Chi L Rev at 545–48 (cited in note 2).

      73   501 US 722 (1991).

      74   Id at 752–54.

      75   Although Coleman formally reserved a ruling on the IAC scenario, that reservation was almost entirely ignored by lower courts. See Primus, 122 Yale L J at 2612 (cited in note 61).

      76   See Megan Raker, Comment, State Prisoners with Federal Claims in Federal Court: When Can a State Prisoner Overcome Procedural Default?, 73 Md L Rev 1173, 1178 (2014).

      77   See Kovarsky, 82 Tulane L Rev at 466 (cited in note 42); Primus, 122 Yale L J at 2609–10 (cited in note 61).

      78   132 S Ct 1309 (2012).

      79   Id at 1315.

      80   See Raker, Comment, 73 Md L Rev at 1178 (cited in note 76). See also Eric M. Freedman, Enforcing the ABA Guidelines in Capital State Post-conviction Proceedings after Martinez and Pinholster, 41 Hofstra L Rev 591, 596 (2014).

      81   See Martinez, 132 S Ct at 1321 (Scalia dissenting).

      82   See note 24.

      83   Martinez, 132 S Ct at 1315 (majority).

      84   See Huq, 81 U Chi L Rev at 601–04 (cited in note 2).

      85   Id at 606.