Habeas Corpus, Relitigation, and Taking Statutes Seriously
“When Congress supplies a constitutionally valid rule of decision, federal courts must follow it.” You wouldn’t think it would be necessary for the Supreme Court of the United States to say that. Everybody knows that. Don’t they? But the Court did find it necessary to say that yesterday in the case of Brown v. Davenport, No. 20-826.
Ervine Davenport was convicted of strangling Annette White to death. His case was thoroughly reviewed by the Michigan appellate courts who ultimately decided that although an error had occurred it had no effect on the outcome. As the Supreme Court has long recognized, ” ‘a defendant is entitled to a fair trial but not a perfect one,’ for there are no perfect trials.” This is the “harmless error” rule.
The general rule in our judicial system is that once a judgment has been reviewed up the appellate chain and affirmed the case is over. With limited exceptions, you can’t go running to another court, especially one that does not have appellate jurisdiction over the court that entered the judgment, and attack the judgment by claiming that the first set of courts got it wrong.
Congress sharply narrowed one of the exceptions in 1996, blocking the lower federal courts from overturning reasonable decisions of state courts merely because they disagree with them. Is there something about the harmless error rule that makes it different so that this statute need not be applied?
The obvious answer is “of course not.” So why did this question even have to come to the Supreme Court?
The old saying “even Homer nods” means that even the great ones in a given field make mistakes occasionally. This is a case of “even Nino nods.”
In yesterday’s opinion of the Court, Justice Gorsuch gives us a thumbnail sketch of the historical ebb and flow of habeas corpus as a means of collaterally attacking criminal judgments and particularly attacking state judgments in federal courts. Justice Kagan, in dissent, challenges both the relevance and accuracy of this history, but it is correct on all material points and illuminates an important point.
Originally, collateral attack was limited to questions of the jurisdiction of the court. If the court had jurisdiction, its decision on the merits of the case was conclusive and precluded collateral attack. In the late nineteenth and early twentieth century, courts began to play fast and loose with what was “jurisdictional” until by 1953 all constitutional issues could be considered whether jurisdictional or not. In that year, the Supreme Court issued a fuzzy opinion in Brown v. Allen which was understood to say that the federal court would decide questions of law without any deference to the state court ruling. However, the same case did give conclusive effect to a state court ruling that the defendant had defaulted a claim by not raising it in the proceeding the state had provided. This is the procedural default rule. Another rule limited filing a second petition after a first one was denied.
By the mid-1960s, even that safeguard was gone. In Fay v. Noia, the Supreme Court blew away the procedural default rule, and in Sanders v. United States it blew away the successive petition rule. It was open season on criminal judgments.
In the 1970s through the early 1990s, the Supreme Court began to pull back the reins. Noting that the statute granted federal courts an equitable power to issue the writ of habeas corpus, the Court held that there were limits beyond which that power ought not be exercised. It brought back the procedural default rule in Wainwright v. Sykes. It brought back the successive petition rule in McCleskey v. Zant. Most relevant to this case, it raised the bar on harmless error in Brecht v. Abrahamson.
In a dubious decision in 1967, Chapman v. California, the Supreme Court decided that constitutional errors could not be judged harmless by the same standard as other errors. A constitutional error would not be considered harmless unless it was harmless beyond a reasonable doubt. Nonconstitutional errors are harmless unless they have a “substantial and injurious effect or influence” on the outcome of the trial.
This distinction has never made any sense. The status of a rule as constitutional or nonconstitutional has no bearing on its importance to the fairness of the trial. The rule excluding evidence obtained in violation of the Fourth Amendment has nothing whatever to do with the fairness of the trial or the reliability of the result. It exists only as a deterrent against future violations. The rule excluding unreliable scientific evidence, on the other hand, has a great deal to do with the fairness of the trial, protecting against conviction on “junk science,” but it is not constitutional.
In 1993, as part of the general retrenchment, the Supreme Court decided in Brecht that a prisoner seeking to overturn a final judgment should have to show something more than just a reasonable doubt that the error he complains of mattered. So the Court adopted the standard for nonconstitutional errors on direct appeal and applied it to all claims on habeas corpus.
In 1996, Congress decided that these piecemeal limitations were not sufficient, and it added additional limitations in the Antiterrorism and Effective Death Penalty Act of 1996. The most controversial of these was a requirement that a federal court could not grant relief on a claim decided against the petitioner on the merits in state court unless the state court decision was (1) contrary to or an unreasonable application of federal law as established by Supreme Court precedent, or (2) based on an unreasonable finding of fact. The first of these, 28 U.S.C. § 2254(d)(1) is the most controversial and is the one involved in this case.
In the 2007 case of Fry v. Pliler, a habeas corpus petitioner made the rather far-fetched claim that AEDPA had supplanted the Brecht rule and restored Chapman as the governing rule for harmless error for constitutional claims in habeas corpus cases. The Court, in an opinion by Justice Scalia, rejected that argument. This is the point of the historical discourse. Congress added an additional requirement for overturning state judgments on federal habeas corpus. It had no intention of lifting the existing requirements. Nothing in the history of text of the act even remotely suggests such a thing.
Near the end of the Fry opinion, Justice Scalia added a remark that was unnecessary to the holding and did not really relate to the issue in the case: “[I]t certainly makes no sense to require formal application of both tests (AEDPA/Chapman and Brecht) when the latter obviously subsumes the former.” By “AEDPA/Chapman” he meant, in a case where the state court on direct appeal applied Chapman and found the error harmless, the question of whether that holding was unreasonable. The remark is irrelevant to the Fry case because the state court in that case did not decide that issue, and § 2254(d)(1) was not implicated.
The question is not at all “certain” or “obvious,” as we will get to in a minute. But does this remark settle an issue that was not in the case? Is it a Supreme Court precedent? No.
Chief Justice John Marshall addressed this problem in his famous opinion in Cohens v. Virginia when he was confronted with his own stray remark from his even more famous opinion in Marbury v. Madison:
It is a maxim, not to be disregarded, that general expressions, in every opinion, are to be taken in connection with the case in which those expressions are used. If they go beyond the case, they may be respected, but ought not to control the judgment in a subsequent suit, when the very point is presented for decision. The reason of this maxim is obvious. The question actually before the court is investigated with care, and considered in its full extent. Other principles which may serve to illustrate it, are considered in their relation to the case decided, but their possible bearing on all other cases is seldom completely investigated.
These “general expressions” that “go beyond the case” are known in law as obiter dicta. As Marshall said, they are not controlling in later cases when the question actually arises. As the Court said more recently in Kirtsaeng v. John Wiley & Sons (2013), “Is the Court having once written dicta calling a tomato a vegetable bound to deny that it is a fruit forever after?”
If Judge A finds a given error harmless beyond a reasonable doubt, and Judge B finds that the same error had substantial and injurious effect, has Judge B necessarily found that Judge A has made an unreasonable application of Supreme Court precedent? That is the crux of the dispute between the majority and dissent in yesterday’s decision.
The first reason why the Fry dictum is not correct is that it fails to take into account the breadth of disagreement among judges in this country. The Supreme Court has said that a decision is reasonable within the meaning of §2254(d)(1) unless “every fairminded jurist” would come to the opposite conclusion. That is an enormously broad standard, but it is a correct interpretation of the statute and the intent behind it. Congress intended to limit federal relitigation of issues already decided by the state court to correction of clear “no question about it” errors.
The second reason is the clause of §2254(d)(1) that limits the federal court to Supreme Court precedent. Congress very deliberately chose to include this limit to stop lower federal courts from overturning state court decisions based on their own precedent, which state courts have no obligation to follow. Only the U.S. Supreme Court can make precedents binding on state courts, and before 1996 some of the federal courts of appeals were getting too big for their britches. They were also often wrong, as it was not unusual for the Supreme Court to resolve disagreements by finding that the state court was right and the federal court was wrong.
Finally, the §2254(d)(1) requirement is the law. Courts cannot simply ignore it.
So, does the federal court have to apply both tests in every harmless error case? No. Almost of the time the federal court can and should decide at the threshold that the state court decision is either correct or at least within the bounds of reasonable disagreement. Decisions so clearly wrong as to fail the test are rare. Federal courts can save themselves a lot of work and delay by just ending most habeas corpus issues right there. And that was the intent of this statute.