Juror egos and juror ids

Samia v. United States and the search for neutral principles in the law of evidence

What’s the most interesting class to take in law school? There are many worthy candidates. Antitrust law is a classic. Federal Indian law is an incredibly deep subject. Patent law, for technology types.

But if forced to choose, I would go with Evidence. Evidence has something for everyone. For those who enjoy complex doctrinal puzzles, the hearsay rules offer unlimited enjoyment. For philosophers, it is hard to beat law school’s version of an epistemology course. For those inclined toward civil rights and social justice, the Rules of Evidence bristle with assumptions worthy of question.

Unfortunately for those interested in the topic, the Supreme Court very rarely hears cases on the Rules of Evidence. Even in the courts of appeals, courts generally resolve evidentiary disputes by citing the discretionary standard of review or finding harmless error.

Happily, however, the Constitution includes its own rule of evidence, which generates Supreme Court litigation at a steady clip. The Sixth Amendment’s Confrontation Clause provides that in “all criminal prosecutions, the accused shall have the right … to be confronted with the witnesses against him.” “Confront” means cross-examine. The Confrontation Clause functions as a kind of constitutional hearsay rule, barring the admission of certain out-of-court statements when the defendant didn’t have the opportunity to cross-examine the declarant.

In March, the Supreme Court will hear argument in Samia v. United States, a Confrontation Clause case concerning the admissibility of confessions in joint trials. Samia illustrates a recurring problem in the law of evidence: judges are forced to craft rules of evidence based on their intuitions of how jurors will react to particular pieces of evidence, despite having no idea whether those intuitions have any basis in reality.

In this post, I will explain the winding path that led to Samia. Then I will explore how one might resolve evidentiary disputes in general, and Samia in particular, without resort to speculation about jury behavior.

The Bruton trilogy

Samia addresses a frequently occurring scenario: a defendant confesses to a crime, is put on trial alongside a co-defendant, and takes the Fifth.

Many crimes are committed by more than one person. Usually, co-defendants are tried together. Defendants are entitled to file motions to sever (that is, for separate trials), and occasionally those motions are granted when the risk of unfairness is too great. But usually these motions are denied. Joint trials are more efficient, prevent witnesses from having to testify multiple times, and avoid the unfairness that arises when one defendant gets a sneak preview of the prosecution’s case and the other defendant doesn’t.

Criminal defendants often confess. Typically, this causes them to plead guilty. Sometimes, however, they press on to trial, urging the jury to ignore their confession. (This rarely works.)

Suppose the confessor insists on a trial. A confession is admissible against the confessor himself. But suppose the confessor takes the Fifth and refuses to testify, and then is tried alongside a co-defendant. Can the confession be used against the co-defendant?

No. This is a paradigmatic Confrontation Clause violation. The co-defendant can’t cross-examine the confessor, so the confessor’s statements can’t be used against the co-defendant.

This creates a problem: the confession, a highly incriminating piece of evidence, is admissible against one defendant but not the other. If the two defendants are on trial together, before the same jury, how should the judge deal with the confession? The Supreme Court has already decided three cases on this issue, in 1968, 1987, and 1998, and understanding Samia requires an understanding of those cases.

Before 1968, judges would admit the confession into evidence against the confessor, but instruct the jury not to consider the confession as evidence against the co-defendant. In principle, this strategy could solve the the Confrontation Clause problem. If—this is a big if—the jurors obey the judge’s instructions and ignore the confessor’s inadmissible confession while deliberating over the co-defendant’s guilt, then the confessor isn’t serving as a witness against the co-defendant.

But the practical reality is that juries don’t always obey the judge’s instructions, and the Supreme Court bowed to this reality in Bruton v. United States, 391 U.S. 123 (1968). Bruton and his co-defendant, Evans, were tried together for armed robbery. Evans confessed orally, saying that both he and Bruton committed the robbery. Evidence of this oral confession was admitted at trial. Evans took the Fifth and refused to testify. The judge told the jury not to consider Evans’ confession against Bruton, but the jury convicted Bruton anyway. The Supreme Court, per Justice Brennan, held that Bruton’s Confrontation Clause right was violated, because the jury couldn’t possibly overcome the temptation of considering Evans’ confession against Bruton, even though the judge told it not to.

Bruton wouldn’t have been decided the same way today. In addition to the current Court being more conservative on criminal procedure than the Court of 1968, Bruton reflects a kind of freewheeling approach to constitutional law that has fallen out of vogue. Today’s Court would say that if the judge instructs the jury not to consider a declarant’s out-of-court statements against a particular defendant, then the declarant isn’t a witness against that defendant and there’s no Confrontation Clause violation, full stop. Maybe there’s a risk the jury will ignore the jury instructions, but calling that risk a Confrontation Clause violation is like finding a Confrontation Clause violation if the jury is exposed to someone shouting about the defendant’s guilt on the courthouse steps. If this seems unfair, the judge can grant a motion to sever, but there is no constitutional issue.

Also, it’s not clear that the Court’s predictions of jury behavior were accurate. Criminals, particularly criminals who refuse to testify, are not trustworthy. Jurors can readily appreciate the unfairness of using Evans’ statements against Bruton when Evans is standing silently at trial, especially if the judge admonishes the jury that it can’t use Evans’ statements against Bruton because of that unfairness. This instruction strikes me as a lot easier to understand than, e.g., the instruction that prior convictions can be used only as impeachment but not as substantive evidence, which is so incomprehensible that defense lawyers typically don’t bother asking for it.

Still, it’s easy to see why Bruton came out the way it did. Why not exclude evidence that Evans incriminated Bruton? Evans confessed his own role; there was zero reason that the jury had to be told he also incriminated Bruton. The only reason the prosecution would oppose excluding this evidence is that it secretly wanted the jury to use Evans’ confession against Bruton. It seemed arbitrary and random not to redact the confession, so the Supreme Court held that admitting the full confession was unconstitutional.

Next up to bat was Richardson v. Marsh, 481 U.S. 200 (1987). Marsh was charged with aiding and abetting a crime committed by her co-defendant, Williams. The question was whether, before the crime happened, Marsh was aware of Williams’ nefarious plans.

Williams confessed to police. Among other things, he stated in his confession that he had announced his plans to commit the crime in a car ride. At the joint trial, Williams’ confession was admitted into evidence, but was scrubbed to omit any reference to Marsh. Williams didn’t testify at the trial, but Marsh did. She acknowledged she had been in the car during the same car ride, but claimed she didn’t hear Williams announce his plans. Marsh was convicted.

According to Marsh, the admission of Williams’ confession at the joint trial violated the Confrontation Clause. Williams confessed that, while in the car, he announced his intention to commit the crimes. In light of Marsh’s acknowledgment that she was in the car, this was by far the strongest evidence that Marsh knew about the crimes in advance. The judge instructed the jury not to use Williams’ confession against Marsh, but we know from Bruton that the jury can’t realistically disregard someone’s confession, right? Williams’ un-cross-examined confession was so devastating to Marsh’s defense that her Confrontation Clause right was violated, or so she argued.

Marsh lost. In a 6-3 opinion, the Supreme Court, per Justice Scalia, distinguished Bruton. The Court pointed out that in Bruton, the confession itself referred to Bruton, while in Marsh, the confession said nothing about Marsh: the jury had to combine the confession with Marsh’s testimony to figure out why the confession was so harmful to Marsh’s case.

That is indeed a difference, but why does this difference matter? The Court offered a creative answer. The Court postulated a theory of juror mind in which the juror id—the psychic agent that instinctively, uncontrollably leaps at incriminating evidence—battles the juror ego—the rational aspect of the psyche that calmly considers jury instructions. In the Court’s view, the victor in this battle turns on the degree of intellectual effort needed to figure out why the un-cross-examined confession is incriminating. If it is easy to figure out why the un-cross-examined confession incriminates the co-defendant, as in Bruton, then the id will overwhelm the ego and cause the hapless juror to ignore the jury instructions. But what if figuring this out requires some mental effort? As the id starts to put the puzzle pieces together, it will trigger psychological tripwire that will cause the ego to swoop in and put a stop to the id’s forbidden thoughts. The ego having saved the day, the juror will obey the jury instructions and use the confession only against the confessor.

On the facts of Marsh, the Court observed that connecting Williams’ confession to Marsh required a bit of thought. Williams didn’t mention Marsh, so the jury had to connect the dot of Williams’ statements about the car ride to the dot of Marsh’s testimony about the car ride. And, in the Court’s view, this dot-connecting required sufficient intellectual effort that the ego would be able to suppress it.

(This is Dall-E’s take on “oil painting of gavel next to brain.”)

This theory is pretty nutty. Is it really that hard for the jury to realize that … whoah … Williams announced his plans in the car and Marsh was in the same car!

Perhaps unconvinced by its own reasoning, the Court drew a second distinction between Marsh and Bruton, which it declared to be “even more significant.” This distinction was that Bruton’s asserted constitutional right was more convenient to vindicate than Marsh’s asserted constitutional right. Bruton was merely asking the judge to exclude Evans’ specific references to Bruton. This request could have been easily implemented via a pre-trial order: The judge could have simply instructed the prosecutor not to elicit any evidence that Evans incriminated Bruton. More generally, written confessions can be redacted before trial to omit references to co-defendants.

By contrast, Marsh argued that Williams’ confession, viewed in light of the rest of the evidence at trial, incriminated her. Before trial, no one knows what the evidence at trial will be, so under Marsh’s position, a judge cannot rule on the Confrontation Clause issue until after all the evidence comes in. This would be inconvenient, as the Supreme Court explained: “The ‘contextual implication’ doctrine articulated by the Court of Appeals would presumably require the trial judge to assess at the end of each trial whether, in light of all of the evidence, a nontestifying codefendant's confession has been so ‘powerfully incriminating’ that a new, separate trial is required for the defendant. This obviously lends itself to manipulation by the defense -- and, even without manipulation, will result in numerous mistrials and appeals.”

It is strange to see an opinion from Justice Scalia that combines extremely speculative predictions about jury psychology with explicit cost-benefit balancing. (Cost-benefit balancing makes sense for Due Process—that is what “due” means—but the Confrontation Clause doesn’t seem to admit of cost-benefit balancing.) But Bruton was similarly free-wheeling, so I suppose he was content to explain why Bruton does not apply on its own terms.

The third case in the pre-Samia trilogy is Gray v. Maryland, 523 U.S. 185 (1998). In Gray, Bell and Gray were tried together for murder. Bell’s confession directly incriminated Gray. At their joint trial, Bell’s written confession was admitted into evidence, but every reference to Gray was replaced with “DELETED.” Gray was convicted.

In the Supreme Court, Gray argued that this was, as a practical matter, the identical case to Bruton because obviously “DELETED” means the co-defendant. Why else would the confession say “DELETED”? Gray argued that Bell’s confession should have either been scrubbed of all references to an accomplice or at least rewritten so it was less obvious that Bell was talking about Gray. Maryland said, no, “DELETED” is good enough.

Along both of Marsh’s axes, Gray lay right on the center-line between Bruton and Marsh. Starting with the intellectual effort axis, Gray’s jury had to think harder than Bruton’s jury (because the confession said “DELETED” rather than explicitly naming the co-defendant, so the jury would have to figure out that “DELETED” meant the co-defendant) but had to think less hard than Marsh’s jury (because the intriguing “DELETED” would naturally pique the jury’s curiosity, whereas the confession in Marsh didn’t mention any co-defendant). Turning to the convenience axis, Gray’s requested relief was less convenient than Bruton’s requested relief (Gray was asking the court to completely rewrite the confession, whereas Bruton was merely asking that the confessor not explicitly identify Bruton), but more convenient than Marsh’s (Gray’s argument didn’t depend on any of the other evidence that came out at trial, whereas Marsh’s did).

A criminal procedure case falling directly in between a Brennan opinion and a Scalia opinion is a classic recipe for a 5-4 decision, and the Supreme Court delivered. Gray won, with the majority consisting of Justices Stevens, O’Connor, Souter, Ginsburg, and Breyer, suggesting the case would also have come out the opposite way today. The majority said that “DELETED” is too juicy and it’s not that hard to rewrite the confession, the dissent said that “DELETED” is less juicy than the person’s actual name and maybe it is that hard.

Deciding cases when you don’t know how to decide them

After a 25-year break, this issue is back. The case is Samia v. United States, scheduled for oral argument on March 29. Samia lies somewhere in between Marsh and Gray, allowing the Supreme Court to resume its binary search for the true Confrontation Clause line.

Samia was accused of being a hitman in an international criminal organization. (The government highlights that Samia expressed interest in “assassinations, but up close and personal,” apparently referred to as “wet work.”) His co-defendant, Stillwell, confessed orally to a DEA agent that while Stillwell was driving a van in the Philippines, Samia was in the passenger seat and committed a murder.

At the joint trial, Stillwell didn’t testify. The government wanted the DEA agent to testify about Stillwell’s confession. The problem was, it was impossible to scrub Stillwell’s confession of any reference to an accomplice. Stillwell’s story was that he was driving a van while someone else committed a murder; you can’t introduce that confession without mentioning the existence of someone else. So to prevent unfairness to Samia, when the DEA agent described Stillwell’s statements about Samia, the DEA agent made elliptical references to an “other person” designed to avoid piquing the jury’s curiosity about who that “other person” might be. For example, the DEA agent testified that Stillwell “described a time when the other person he was with pulled the trigger on that woman in a van that he and Mr. Stillwell was driving.” The judge instructed the jury to use Stillwell’s confession against Stillwell only.

According to Samia, however, the rest of the trial evidence made it obvious that the “other person” referred to in the confession was Samia. As such, according to Samia, the jury inevitably used Stillwell’s confession against Samia, despite the judge’s instruction not to, yielding a Confrontation Clause violation.

As noted above, this case lies in between Marsh and Gray. On the intellectual effort axis, Samia’s jury had to think harder than Gray’s jury (because the Samia jurors had to mentally combine Stillwell’s confession with the other evidence at trial, whereas the Gray jurors could easily figure out that “DELETED” meant Gray), but had to think less hard than Marsh’s jury (because Stillwell’s confession contained a reference to a co-defendant, whereas the confession in Marsh didn’t mention any co-defendant). As to the convenience axis, Samia’s case is more similar to Marsh. As in Marsh, but unlike in Gray, there was nothing the judge could do pre-trial to cure the Confrontation Clause violation, other than exclude the confession or sever the trials.

So how should the Court decide Samia?

Well, one way is for the Justices to introspect about whether a jury, on these facts, could follow the jury instructions and overcome their temptation to use Stillwell’s confession against Samia. This approach is dissatisfactory. No one has any idea whether the additional increment of intellectual effort here relative to Gray is enough to withstand a juror’s burning desire to disobey the jury instructions.

To me, the Court’s central mission in Samia is to decide the case without making wild guesses as to jury psychology. But how is this possible?

One approach is a kind of Talmudic approach to Supreme Court precedent. Like many lawyers, I went to religious schools growing up, in which we regularly studied the Talmud. Frequently, we would encounter multiple passages that, at first glance, pointed in opposite directions. However, our premise was that the Talmud was the word of God, so we did not have the option of finding one passage more persuasive and limiting the other passage to its facts. Instead we had to diligently try to reconcile the passages while affording both passages equal respect.

Adopting a similar approach, we can treat Marsh and Gray as sacred texts, deserving of equal weight in the jurisprudential calculus, and focus exclusively on whether Samia is factually closer to Marsh or Gray. Under this approach, the government should win. If we conceptualize Marsh, Gray, and Samia as points on a Cartesian plane, with one axis representing intellectual effort and the other representing convenience, Samia is geometrically closer to Marsh than Gray. As mentioned above, Samia lies in between Marsh and Gray on the intellectual effort axis and is equal to Marsh on the convenience axis, so Samia is closer to Marsh overall. Note that resolving the case this way does not require any speculation about jury behavior, it simply requires mechanically comparing Samia to existing case law.

It is easy to make fun of this approach, but to me, it is pretty good. Aren’t we often told that the purpose of precedent is to liquidate the meaning of vague constitutional texts and narrow the zone of disagreement? That we can’t relitigate Marbury v. Madison in every single case? This approach offers a straightforward and determinate way of reaching a legal answer, and I could not confidently say that introducing additional degrees of freedom into the legal analysis would yield an answer that is somehow “better.”

But suppose we are a little more inquisitive. Suppose we diligently want to get to the correct legal answer. But suppose there is no way to obtain a correct answer to the ultra-speculative question of whether juries would be tempted to ignore the jury instructions on these facts or not. What should we do?

We should use second-order “who decides” reasoning. Samia is essentially arguing that his trial should have been severed from Stillwell’s trial. Someone has to decide that issue. One option is the trial judge, in the context of a discretionary motion to sever. Another option is the Supreme Court, in the name of the Constitution.

Neither trial judges nor Supreme Court Justices can reliably predict jury behavior. But if I had to pick a decisionmaker, it would be trial judges, who at least have some experience with juries. Only two of the nine Justices (Justice Sotomayor and Justice Jackson) have ever served as trial judges, in both cases for relatively short periods.

Also, the stakes of trial judges’ evidentiary decisions are lower. Supreme Court rulings interpreting the Constitution are binding on all state and federal courts nationwide. Trial court rulings are not precedential anywhere, and even when they are affirmed on appeal, appellate courts apply abuse-of-discretion review, implying that subsequent trial courts can reach inconsistent conclusions and still be affirmed. Also, Supreme Court rulings last forever. It’s unlikely that decisions in this obscure corner of the law will ever be overruled. So if Samia wins, then all judges, for the rest of time, will face a mandatory rule of excluding confessions on similar facts. By contrast, nonconstitutional evidentiary rulings, even when affirmed by appellate courts, can be overturned not only by the legislature, but also by the judicial panels that annually produce updated rules of evidence.

In my view it is hard to see a justification for the Supreme Court declaring a mandatory constitutional severance rule. Given the extreme degree of uncertainty here on whether juries will obey their instructions, there is something to be said for case-by-case decisionmaking, as opposed to a universal rule across both place and time.

I recognize that similar arguments can be made against the Supreme Court enforcing the Constitution in all cases. Supreme Court rulings are always permanent, national, and imposed by Supreme Court Justices. But typical arguments in favor of muscular judicial review don’t apply. The Supreme Court may have strong institutional reasons to subject legislative or executive action to searching judicial review. But Supreme Court Justices and lower-court judges are on the same team. It is legal lore that trial judges are better-positioned to make evidentiary rulings than appellate judges—that’s why abuse-of-discretion review applies—and there is no reason that principle should change merely because a constitutional question is purportedly on the table. Of course, the Supreme Court is duty-bound to be faithful to the text of the Constitution, no matter the consequences, but here it doesn’t seem that a decision in either direction would promote fidelity to the constitutional text.

As I have grown up as a lawyer, I have learned more and more about what I don’t know. We should know what we don’t know, and apply rules designed for cases where we don’t know, rather than pretending we know.

Comments are welcome! In next week’s post, I will discuss PredictIt’s litigation against the CFTC.