Criminal Procedure: Wilson v. Sellers

By Kathy Swedlow

Published October 2017

Click here for a print-friendly PDF of this article.

How Should a Federal Habeas Court Evaluate the Highest State Court’s Summary Denial Order When a Lower State Court Has Issued a Reasoned Opinion on the Merits?


After being convicted of murder and sentenced to die, Marion Wilson claimed that his trial lawyers were ineffective. A Georgia Superior Court considered and rejected those claims in a “reasoned opinion,” and the Georgia Supreme Court affirmed in a “summary order.” Now, on federal habeas corpus review, the federal courts must decide which “adjudication on the merits” to review: the Superior Court’s reasoned opinion or the Supreme Court’s summary order.


Did the Court’s decision in Harrington v. Richter, 562 U.S. 86 (2011), silently abrogate the presumption set forth in Ylst v. Nunnemaker, 501 U.S. 797 (1991)—that a federal court sitting in habeas proceedings should “look through” a summary state court ruling to review the last reasoned decision—as a slim majority of the en banc Eleventh Circuit held in this case, despite the agreement of both parties that the Ylst presumption should continue to apply?


While shopping at a local store, Marion Wilson and a second man, Robert Butts, approached an off-duty correctional officer, Donovan Parks, and asked for a ride. Parks agreed to help. Wilson later told police that Butts, sitting in the front seat of the car with Parks, pulled out a sawed-off shotgun, ordered Parks from the car, then shot and killed him. Subsequently, three other prisoners claimed that Butts confessed that he shot Parks. Butts denied killing Parks.

At his trial, Wilson claimed that he was merely present during the crime and had reduced culpability relative to Butts. In support of his defense, Wilson attempted to introduce the hearsay testimony of the three prisoners to whom Butts had confessed, arguing that a conspiracy exception to hearsay supported admission of the testimony. The trial court disagreed, finding that any conspiracy between Wilson and Butts ended with Wilson’s arrest. And, despite the fact that the Georgia Supreme Court had an established procedure to address the admission of hearsay statements offered in support of “a claim that someone other than the defendant is responsible for the crimes being tried,” Wilson’s trial counsel “did not utilize the procedures set forth [under Georgia law] and did not obtain a ruling from the trial court evidencing its consideration of the proffered hearsay evidence under [Georgia law].” Wilson v. State, 271 Ga. 811, 815, 525 S.E.2d 339 (1999). Without hearing from the three prisoners, the jury convicted Wilson of first degree murder.

At the penalty phase, Wilson’s attorneys presented a “paltry” case for life, which included testimony of his mother and a forensic psychologist; the latter relied “mostly on records obtained by trial counsel,” the petitioner wrote. The jury also heard testimony from a defense investigator, who described his conversations with two of the prisoners to whom Butts had confessed. In response, the prosecution presented 22 witnesses “who described Wilson’s purported gang involvement and history of juvenile impulsiveness and violence.” After deliberating a short while, the jury sentenced Wilson to death.

At his trial, Wilson was represented by two lawyers, neither of whom had prior capital trial experience or training.

The first attorney, Thomas O’Donnell, failed to disclose that he had accepted a position as a Special Assistant Attorney General. When O’Donnell did disclose his work for the state, posttrial, Wilson, “with some equivocation,” agreed to have O’Donnell continue with the appeal. However, after the Attorney General’s office explained that “the potential for a conflict is too much risk to take in a death penalty action,” O’Donnell was removed. The second attorney, Jon Carr, was disbarred in 2007 after being convicted of child molestation. Both O’Donnell and Carr had spouses who worked in corrections, and O’Donnell’s wife knew the victim; O’Donnell later admitted to feeling “pressure about the case” from the corrections community. Wilson also alleges that “[b]ecause each lawyer believed the other was responsible for the mitigation investigation, no meaningful investigation was done.”

On direct appeal, the Georgia Supreme Court affirmed and the Supreme Court denied certiorari. Wilson v. State, 271 Ga. 811, 525 S.E.2d 339 (1999), cert. denied, 531 U.S. 838 (2000).

Wilson filed a habeas petition in the Butts County Superior Court, alleging that O’Donnell and Carr rendered ineffective assistance of counsel for failing to investigate, develop, and present significant mitigation evidence in support of a life sentence. After a hearing where Wilson offered a much more complete picture of his difficult childhood and adolescence, as well as his mental health problems and cognitive impairments, the court denied relief in an unreported, written opinion. Wilson then sought a certificate of probable cause to appeal to the Georgia Supreme Court, but in a summary order, that court declined to hear the case:

Upon consideration of the Application for Certificate of Probable Cause to appeal the denial of habeas corpus, it is ordered that it hereby be denied. All the Justices Concur.

Wilson next turned to the federal courts, filing a federal habeas petition. The district court found that O’Donnell’s and Carr’s performance was “difficult to defend,” but that Wilson could not show prejudice for his ineffectiveness claims. In reviewing Wilson’s claims, the district court relied on the two state court opinions: the Georgia Supreme Court’s opinion denying relief on direct appeal, and the Superior Court’s opinion denying state habeas relief. Wilson v. Humphrey, 2013 WL 6795024 (M.D. Ga., December 19, 2013).

After the district court issued a certificate of appealability, a panel of the Eleventh Circuit affirmed. Wilson v. Warden, 774 F.3d 671 (11th Cir. 2014). The panel began its analysis by explaining that, instead of reviewing the Superior Court’s opinion denying relief, it would instead review the Georgia Supreme Court’s summary order to determine “whether there was any reasonable basis for [that court] to deny relief.” Id. at 678 (citations and internal quotations omitted). As the panel explained, Harrington v. Richter, a 2011 decision from the Supreme Court, mandated review of the “final decision on the merits,” that is, the Georgia Supreme Court’s summary order, even though the order offered no reasoning for the court’s ruling. Thus, in denying relief, the Eleventh Circuit repeatedly referred to what the Georgia Supreme Court “could have” decided when it denied relief.

Wilson moved for rehearing en banc, arguing that the panel erred when it reviewed the Georgia Supreme Court’s summary denial and that it should have examined the Superior Court’s opinion instead. Georgia responded, arguing the Georgia Supreme Court’s summary order should control. Rehearing was granted. Georgia then “changed its position” and argued that the en banc court ought to rely on the Superior Court’s opinion. As Wilson and Georgia were seemingly in agreement on this issue, the court then appointed amicus counsel to argue the opposite position.

After briefing, the en banc court denied relief, basing its ruling on two conclusions.

First, the en banc court held that when the Georgia Supreme Court denies a certificate of probable cause, its denial order is an “adjudication on the merits.” The en banc court based this determination on Georgia Supreme Court Rule 36, which directs that the court “will” grant certiorari when the appeal presents “arguable merit.” By contrast, the en banc court explained that other states have a discretionary review process, “similar to [the Supreme Court’s] certiorari review.” The en banc court explained in those other jurisdictions, a court makes a decision to accept the case depending, for example, on “the interests of justice” or “considerations other than the merits of the appeal.” Wilson v. Warden, 834 F.3d 1227, 1232–1236 (11th Cir. 2016).

Second, the en banc court held that the Antiterrorism and Effective Death Penalty Act (AEDPA) and Richter require that a federal court identify and rely upon the “last adjudication on the merits” and not necessarily a court opinion, explaining the reason for the denial. Rejecting Wilson’s argument that the holding in Richter depended on its unique procedural history—it involved only a summary denial from the California Supreme Court, and there was no lower court opinion to “look through” to—the en banc court cautioned that neither the AEDPA nor Richter suggested that there should be “two divergent analytical modes” for review, that is, “one for when there is no previous reasoned decision below and another for when there is.” Wilson, 834 F.3d at 1236–1242.

Five of the 11 judges of the en banc court dissented, in two separate dissents. In the first, the dissenters began by candidly acknowledging that the best they could do “is predict which line of authority the Supreme Court will use to decide” this issue. Wilson, 834 F.3d at 1242 (dissenting opinion). This is because, in the dissenters’ view, both Ylst and Richter were silent on this point, and “silence is a legal wash.” But, citing Justice Holmes’s adage that “the life of the law has not been logic, it has been experience,” the dissenters claimed that there is strong indication that the Ylst “look through” procedure should not be abandoned after the AEDPA. In support, they noted that in Richter’s companion case, Premo v. Moore, the Supreme Court “did not explicitly state that it was looking through to the last reasoned state court decision, but that is what [it] seemed to do.” Wilson, 834 F.3d at 1243. The dissenters also noted Justices Ginsburg’s and Kagan’s concurrence in the denial of certiorari in Hittson v. Chatman, which criticized the Eleventh Circuit for “discarding Ylst,” and noted that “the circuits to have considered the look-through issue limit Richter to situations where there is no reasoned decision by any state court.” Wilson, 834 F.3d at 1244–1245. (The second dissent echoed many of the same arguments.)

The Supreme Court granted certiorari on February 27, 2017, and also appointed amicus counsel to brief and argue in support of the judgment below—an apparent necessity because the State had sided with Wilson before the en banc court. Several weeks later, the State submitted a letter to the Court Clerk, explaining that it would defend the Eleventh Circuit’s en banc ruling. The Supreme Court then withdrew that portion of its order, appointing amicus counsel.


Under AEDPA, a federal habeas court reviewing a state court’s decision must first determine whether that decision “adjudicated” the habeas petitioner’s federal constitutional claim “on the merits.” 28 U.S.C. § 2254(d).

If the state court has not adjudicated the claim on the merits, then the federal court is free to consider the claim de novo. However, if the state court has adjudicated the claim on the merits, a federal court must employ a highly deferential standard of review and will be barred from granting relief unless the state court’s adjudication was “contrary to” or “involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States…” 28 U.S.C. § 2254(d)(1). “This task is straightforward when the last state court to decide a claim has issued an opinion explaining its decision.” Hittson v. Chapman, 135 S. Ct. 2126 (2015) (Justices Ruth Bader Ginsburg and Elena Kagan, concurring in the denial of certiorari).

In 1991, the Court explained how federal courts should handle the more “challenging circumstance” presented when the last state court issues only a summary order. Id. at 2127. In those circumstances, “where there has been one reasoned state judgment rejecting a federal claim,” a federal court should assume that “later unexplained orders upholding that judgment or rejecting the same claim rest upon the same ground.” Ylst v. Nunnemaker, 501 U.S. 797, 803 (1991). Because unexplained orders “say nothing,” the federal court should instead “look through them to the last reasoned decision.” Id. at 804.

Ylst was decided prior to the enactment of the AEDPA. Post-AEDPA, the Court has held that AEDPA includes a rebuttable presumption that a state court’s silence on the merits of a federal constitutional claim is an adjudication on the merits. Harrington v. Richter, 562 U.S. 86 (2011). This rebuttable presumption applies when the state court denies relief in a one-sentence summary order and when the state court expressly rejects some claims, but does not address others. Id.; Johnson v. Tara Williams, 568 U.S. 289 (2013). In these situations, the habeas court then engages in a hypothetical inquiry to ask “what arguments theories…could have supported the state court’s decision[,] and then it must ask whether it is possible fair-minded jurists could disagree that those arguments or theories are inconsistent with the holding in a prior decision of [the Supreme Court].” Richter, 562 U.S. at 102.

In June 2015 and foreshadowing Wilson’s case, the Court denied certiorari in a Georgia capital case where the Eleventh Circuit had “discard[ed]” Ylst in favor of Richter. Hittson, 135 S. Ct. at 2127. Concurring in the denial of certiorari, Justices Ginsburg and Kagan explained that it was irrelevant that the Georgia Supreme Court exercised discretion to review a case or not. Id. at 2128. In addition, the justices explained that “Richter’s hypothetical inquiry was necessary because no state court opinion explained the reasons relief had been denied.” Id. at 2127 (emphasis in original; internal citations and quotations omitted). However, despite the Eleventh Circuit’s “clear[] err[or] in declining to apply Ylst,” the justices concurred in the denial of certiorari as they were “convinced that the Eleventh Circuit would have reached the same conclusion had it properly applied Ylst.” Id. at 2128.

Identifying the State’s position: At the outset, it is important to identify which side of the argument the State of Georgia now takes, as it has taken multiple contrary positions in these habeas proceedings. First, before the Eleventh Circuit, and in response to Wilson’s petition for rehearing en banc, the State argued that the panel properly relied on the Georgia Supreme Court’s opinion. Second, after rehearing was granted, the State “changed its position” and argued that the Superior Court opinion should control. (Due to this change of position, the Eleventh Circuit appointed amicus counsel to brief and argue in favor of using the Georgia Supreme Court’s summary order.) Third, in response to Wilson’s petition for a writ of certiorari, the State explained that it did not oppose review, given the split that had developed in the circuits after the en banc ruling. The State’s neutrality presumably led to the filing of a brief by amicus counsel, arguing—as he had in the Eleventh Circuit—that Georgia Supreme Court’s summary order should control. Fourth and finally, after certiorari was granted and the Supreme Court re-appointed amicus counsel, the State filed a letter with the Clerk, explaining that it would go back to its earlier position, that is that “deference under Section 2254(d) should be given to the last state court decision on the merits, even when that adjudication is unexplained.” Given the State’s shifting arguments, the certiorari question—which refers to “the agreement of both parties that the Ylst presumption should continue to apply”—is no longer correct. However, Wilson’s presumption that the State agreed with his position is, to say the least, understandable.

Wilson’s arguments: Wilson presents three basic arguments. First, he criticizes the en banc opinion below, claiming that it ignores “the unique context out of which Richter emerged.” Wilson begins this argument by explaining the “paradigm shift” imposed by the AEDPA: it “move[d] the state court’s decision on the merits…from the margin to the center of federal habeas review,” and it limited the power of the federal courts to grant relief. As such, Wilson claims, the federal courts have a significant need to know “what the state court knew and did” in order to correctly apply Section 2254(d). Against that backdrop, Wilson emphasizes the “idiosyncratic” California procedure that led to Richter: that initial state habeas petitions be filed in the appellate courts, and not the trial courts. This, Wilson explains, made California a “uniquely prolific producer of truly ‘summary’ dispositions,” and so “the issue before the Court was an all-or-nothing proposition.” In other words, the Court had to find that the summary denial procedure in Richter was an “adjudication on the merits,” because otherwise every California habeas case would have to be reviewed de novo.

In Wilson’s view, Richter should continue to apply in Richter-like circumstances, and Ylst’s “look through” procedure should continue to be applied in non-Richter cases. As for the en banc court’s concern about then having “two divergent analytical modes” of review, Wilson cites the many habeas corpus cases where the federal courts have accommodated both modes while still keeping focus on what happened in the state court, as the AEDPA requires. Indeed, Wilson observes that if having “two such modes were truly a problem demanding the solution imposed below, surely this Court would have noticed and said so by now.”

Second, Wilson argues that Ylst’s “look through” procedure is the best procedure to employ in a state like Georgia, as it honors the “substantial time and resources” the state invests “in the taking of evidence and the production of reasoned lower court opinions.” Wilson cautions that to apply Richter and its resulting hypotheticals would detract from “what the state court knew and did”—and possibly even exclude the state court’s actual analysis from federal habeas review.

Third, Wilson argues that applying Richter to all habeas cases will be harmful, and uses his case as an example. As Wilson explains, the state Superior Court rejected his ineffectiveness claims because it found—without explaining precisely why—that his witnesses’ testimony would have been inadmissible and cumulative, and because he did not show prejudice. But the en banc court went further and speculated that the Georgia Supreme Court could have denied relief because it thought that Wilson’s mitigation was “double-edged” or that his expert used flawed methodology. In Wilson’s view, this is what will happen if Richter is applied to all cases: as long as a court can create a hypothetical reason for the state court’s denial of relief, it will not matter if those claims were actually denied unreasonably.

The State’s arguments: The State responds with three primary arguments of its own. First, it argues that the text of the AEDPA supports the approach taken by the en banc court below. In its view, the references within Section 2254 to “a decision” means that there is no requirement that the federal court only give deference to an “opinion.” See 28 U.S.C. Section 2254(d)(1) and (2).

While the State concedes that nothing in the AEDPA prevents a federal court from examining a lower court opinion, it also claims that in a case like this one, the only reason for a federal court to do so would be to answer Richter’s hypothetical inquiry. “Indeed, if that opinion supplies a reasonable basis in law and fact that could have supported the later summary decision, the federal court’s inquiry ends there.” But if the lower court’s opinion “unreasonably applied or determined the relevant law or facts,” the State contends that the federal must then engage in Richter’s hypothetical inquiry to determine whether “another reasonable basis could have supported the last state-court merits decision.”

Second, the state offers policy rationales for its interpretation of Richter: that maintaining the “look through” procedure would “impose an impermissible opinion-writing standard” on the state courts and would put “federal courts back in the paternalistic relationship to state courts AEDPA was designed to end.” In the State’s view, a victory for Wilson would mean that state courts “would either have to do away with summary dispositions altogether” or the state courts would have to “at least provide an additional statement indicating that the court does not necessarily agree with the reasoning of the court below.” This would place “the federal court[s] back in the kind of tutelary relation to the state courts that the AEDPA was designed to end” (internal quotation and citation omitted).

Third, the State claims that since the federal courts do not assume that summary affirmances in federal cases adopt the opinion below, it makes “good sense” for the Court to assume the same is true in the state courts. Moreover, the State contends that it would show a “striking lack of respect” for the state courts, if a federal court was permitted to treat a state court summary order differently than it would treat its own.


For followers of habeas corpus law, the arguments in this case present a strange role reversal.

The State argues that to presume that the Georgia Supreme Court agreed with the Superior Court shows a “lack of respect” for the state courts. However, at the same time, the State doesn’t seem to find it equally (or even more) disrespectful for a federal court to hypothesize about the Georgia Supreme Court’s possible reasoning. In contrast, Wilson argues the opposite: that Richter should be restricted to cases where there is no opinion to look through to and that the federal courts should properly focus on “what the state court knew and did.”

The Court here is faced with the impossible task of establishing rules for interpreting silence. During oral argument in Richter, Justice Anthony Kennedy asked about the difference between the California Supreme Court’s summary denial orders and those that include the additional language, “on the merits.” In response, the deputy attorney general explained that “local practice,” as “understood by the court’s litigants,” is that a summary denial from the California Supreme Court is a denial on the merits, even if the denial order doesn’t reference the merits. That response illustrates the problem here: with 50 different states and 50 different examples of “local practice,” how can the Court plausibly craft a workable rule for how the federal courts should interpret state court silence? Even if local practice in a jurisdiction could be ascertained, what if it changes as the composition of a court changes? What if there is no established local practice?

The best answer to these questions might be Wilson’s, which seems to be one shared by Justices Ginsburg and Kagan in Hittson: “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 (citations and internal quotation removed).