Opinion for the Court filed by Circuit Judge STEPHEN F. WILLIAMS.
Concurring opinion filed by Circuit Judge RANDOLPH.
Opinion dissenting in part filed by Circuit Judge SILBERMAN.
On May 2, 1989 a federal jury convicted Michael Pryce, Calvin Thomas, Donovan Antonio, and Nathaniel Gaskins of committing various drug and firearm crimes.1
On December 20, 1988 the police raided Marguerite Briscoe’s apartment in Southwest Washington. There they found Thomas and Pryce, along with much crack, a little over $2000 in cash, and a nine-millimeter pistol. They also found, among other people, Briscoe’s son Reginald Chandler and Reginald’s cousin Anthony Chandler. Several weeks later, the police arrested the other two defendants (Antonio and Gaskins) in a nearby apartment and charged them with involvement in the drug ring.
At trial, the government offered as its principal witnesses the police officers who conducted the December 20 raid, and — for an insider’s view of the drug ring — Briscoe and the two Chandlers. (Reginald, and perhaps Anthony, testified under a grant of immunity from the government.) According to the insiders, Thomas, Pryce and Antonio began living in the apartment and using it as a drug distribution center around the beginning of December, at which time Briscoe moved out. Business grew. According to Reginald, Gaskins arrived about a week or ten days later “to work with” Antonio, and then left about a week after that. The testimony generally suggested that Thomas and Pryce were the leaders of the operation, while Gaskins and Antonio were, by comparison, bit players with look-out and perhaps enforcement roles.
Our first — and most important — issue concerns the trial court’s decision to prohibit defense counsel from cross-examining Anthony Chandler on his past hallucinations. Just before his cross-examination, Gaskins’s lawyer told the court that he had access to a psychiatric report, dated September 26, 1988, stating that Anthony had been seeing and hearing nonexistent events. After some preliminary discussion, the court and the lawyer had this colloquy:
THE COURT: I would say this. I will let you ask him if he suffered any during the relevant time frame, but you may not go into that report or anything he may have told a psychiatrist.
MR. HAND: Very well.
THE COURT: I think, as I said, you can test the credibility of the witness and you can test the ability of the witness to observe. So you can ask him, if you want to take that risk, because it is risky.
MR. HAND: Well, may'I ask him, “Do you suffer from”—
THE COURT: No.
MR. HAND: “Do you ever hear things and see things that aren’t really there?” THE COURT: Only during the time frame that he has discussed, December 1 through December 20. You can run the risk if you wish.
MR. HAND: Very well.
5/4 Tr. at 106-07. The trial court effectively ruled that any cross-examination of Anthony on his mental condition would have to begin and end with questions about that condition as it existed in December. The court thus apparently barred defense counsel from asking any questions about Chandler’s hallucinations in September as a foundation for questions about his condition in December, much less as a basis for impeaching his responses about December. The court gave no explanation for its ruling.
This restriction was an abuse of the trial court’s discretion to limit cross-examination on matters affecting credibility, see, e.g., United States v. Partin, 493 F.2d 750, 762-64 (5th Cir.1974); United States v. Society of Independent Gasoline Marketers of America, 624 F.2d 461, 467-69 (4th Cir. 1979), and violated the confrontation clause of the sixth amendment, see Delaware v. Van Arsdall, 475 U.S. 673, 679-80, 106 S.Ct. 1431, 1435-36, 89 L.Ed.2d 674 (1986);
The government argues that appellants failed to “establish a foundation” for their claim because they decided not to cross-examine Anthony Chandler on his hallucinations. At first blush, that argument seems absurdly unfair to criminal defendants, since it would require them, in cross-examining a prosecution witness, to ask the clincher question first and, upon receiving an unfavorable answer, drop the inquiry in silence. As the trial court itself admitted, that approach is quite “risky” — it could easily leave defense counsel in front of the jury holding a popped balloon, looking at once ridiculous and obnoxious.
The government relies for its foundation argument on Luce v. United States, 469 U.S. 38, 105 S.Ct. 460, 83 L.Ed.2d 443 (1984). The defendant in Luce had asked the trial court for an in limine decision barring the government from cross-examining him on his prior convictions if he took the stand. The court denied the motion, and the defendant did not take the stand. Upon his conviction, he challenged the court's evidentiary decision under Federal Rule of Evidence 609(a)(1). The Supreme Court held that Luce’s decision not to testify foreclosed him from bringing this claim on appeal.
We have found no case considering whether Luce extends to the general problem at issue here: whether during cross-examination a party must ask “risky” question X in order to preserve his right to challenge a court’s decision to keep him from asking related question Y. And the rationale of Luce appears inapplicable. First, that case concerned an in limine ruling that came well before the relevant witness was to have taken the stand and delivered his testimony. To the extent that the witness’s actual testimony (or other developments in the trial) could have affected the factual context on which the ruling depended, the trial judge might have changed his ruling at the appropriate time. Id. at 41-42, 105 S.Ct. at 463. That, of course, is not the case here: at the time of the trial court’s ruling, Anthony had already testified for an extended period, and
More fundamentally, the Luce Court noted that “[bjecause an accused’s decision whether to testify seldom turns on the resolution of one factor ... a reviewing court cannot assume that the adverse ruling motivated a defendant’s decision not to testify.” Id. at 42, 105 S.Ct. at 463 (internal quotations omitted). And it expressed concern that if non-testifying defendants could appeal such rulings, they could use in limine motions “solely to ‘plant’ reversible error in the event of conviction.” Id. Absent the Luce rule, a defendant who never intended to testify could pursue such motions as a no-lose strategy. If his motion were rejected, he could appeal; if successful, he could generally choose not to testify anyway. See id. at 41 n. 5, 105 S.Ct. at 463 n. 5. By contrast, if the court had permitted Gaskins’s counsel to use the psychiatric report, it is hard to imagine that he would not have done so. A temptation to “plant reversible error” appears farfetched here.
The Court also noted that where a demonstrably erroneous evidentiary ruling arguably keeps a defendant from testifying, “the reviewing court would still face the question of harmless error.” Id. at 42, 105 S.Ct. at 463; see also id. at 43, 105 S.Ct. at 464 (Brennan, J., concurring). Not knowing what the testimony might have looked like, a court would be hard pressed to assess, first, whether it would have done the defendant any good, and, second, whether the other side’s cross-examination would have been effective. While the unconducted cross-examination here is an unknown, it might well have led the jury to reject Anthony Chandler’s testimony. To be sure, if appellants had gone ahead and asked Anthony about his psychiatric status in December, he might have immediately admitted to hallucinating during the relevant period. That outcome might have rendered the trial court’s error harmless, though only if the jury convicted anyway. This possibility alone, however, does not outweigh the unfairness to these appellants of requiring them to pursue potentially self-destructive trial tactics simply to preserve their constitutional rights.
Having found that the trial court committed error and that appellants have the right to appeal it, we must now consider whether the government’s failure to assert harmless error requires us to reverse the convictions even of defendants who suffered no prejudice, or (alternatively) whether it should alter the harmless error inquiry. I here state my own approach to the matter; Judge Randolph’s view, leading him to concur in the result, and Judge Silberman’s, leading him to dissent in part, follow in separate opinions.
Where a court analyzes the harmless error issue wholly on its own initiative, it assumes burdens normally shouldered by government and defense counsel. This drain on judicial resources inevitably causes delay for parties in other cases. See United States v. Giovannetti, 928 F.2d 225 (7th Cir.1991). More important, where the case is at all close, defense counsel’s lack of opportunity to answer potential harmless error arguments may lead the court to miss an angle that would have shown the error to have been prejudicial. On the other hand, where the trial error plainly did not contribute to a defendant’s conviction, automatic reversal inflicts on the public the costs of a needless retrial and on other litigants the resulting delays. Reconciling these competing concerns, the Giovannetti court recognized its discretion to proceed with the harmless error analysis despite the government’s failure to raise the issue, and identified the following factors as controlling its exercise of that discretion:
... [1] the length and complexity of the record, [2] whether the harmlessness of the error or errors found is certain or debatable, and [3] whether a reversal will result in protracted, costly, and ultimately futile proceedings in the district court.
I agree with the general approach of Giovcmnetti. Only if one adopts an absolutist approach to the adversary system can one contend that courts must never address unargued issues, no matter how obvious their proper resolution may be. Certainly the Supreme Court rejects such an approach. In United States v. Karo, 468 U.S. 705, 104 S.Ct. 3296, 82 L.Ed.2d 530 (1984), for instance, after holding that certain evidence supporting a search warrant was unconstitutionally gathered, the Court went on without the benefit of briefing to decide that the violation was harmless in light of untainted evidence that independently established probable cause. Id. at 719-21, 104 S.Ct. at 3305-06; see also id. at 736, 104 S.Ct. at 3314 (opinion of Stevens, J., dissenting) (noting that the issue “was not raised in the petition for certiorari and has not been briefed by the parties”); see also Teague v. Lane, 489 U.S. 288, 109 S.Ct. 1060, 1069, 103 L.Ed.2d 334 (1989) (addressing a retroactivity issue that was addressed only in a single amicus brief). Moreover, in conducting a harmless error analysis without briefing, we are not plucking issues from thin air. As in Karo, the parties here argued the main issue (the propriety of the district court’s ruling) and simply failed to address its likely impact on the final result.
To be sure, a court should normally conduct the harmless error inquiry on its own initiative only where the relevant portions of the record are reasonably short and straightforward. Moreover, where a court does conduct the inquiry on its own, it should err on the side of the criminal defendant. In the normal case, where the government does raise the issue, the relevant standard for confrontation clause violations is “whether, assuming that the damaging potential of the cross-examination were fully realized, a reviewing court might nonetheless say that the error was harmless beyond a reasonable doubt. ” Delaware v. Van Arsdall, 475 U.S. at 684, 106 S.Ct. at 1438 (emphasis added). Where the government does not raise the harmless error issue, I would deem errors “harmless” only where satisfaction of that standard is beyond serious debate. See Giovannetti, 928 F.2d at 226-27. Here, as the record is neither long nor complicated, I think the inquiry suitable.
Judge Randolph concurs in the analysis of the facts below, although, as I understand his position, he regards as surplus-age the references to whether a finding of harmlessness is indisputable.
As to three defendants (all but Gas-kins), the error was indisputably harmless. It obviously did not harm defendant Antonio, for instance, since Anthony Chandler never once mentioned him during his testimony. One might argue that Anthony’s testimony bolstered the testimony of government witnesses Reginald Chandler and Marguerite Briscoe — who did finger Antonio — because all three testimonies were reasonably consistent with one another. Yet inferring from this that the error prejudiced Antonio requires a caution so extreme as almost to preclude finding any error harmless. First, the testimony of the police officers was also consistent with that of Reginald and Briscoe. More fundamentally, the net effect of Anthony’s testimony may well have been beneficial to Antonio, as his total omission of Antonio’s name would have tended (if anything) to make jurors question whether Antonio really was part of the drug gang.
Nor can we imagine a serious argument for reversing the convictions of appellants Pryce and Thomas.5 As the arresting offi
The issue becomes harder when we consider whether the trial court’s error contributed to appellant Gaskins’s conviction. One could argue that because Briscoe and Reginald Chandler both implicated him in their testimony, Anthony Chandler’s testimony was merely cumulative — and harmless. Perhaps so. But as the matter is open to serious debate, we reverse Gas-kins’s conviction.
First, Briscoe hardly ever mentioned Gaskins in her testimony. On direct, she made several spot references to his presence in the apartment, see 5/3 Tr. at 5, 8, 12, 15-16, but her most damaging testimony, if it can be called that, is that Gaskins once stood by looking at a crack transaction but “didn’t do anything.” Id. at 15-16. Only once did she accuse him of clearly criminal activity. On cross-examination, after she denied that her relatives — the Chandler cousins — ever dealt drugs in the apartment, a defense lawyer probed further, hoping for a different answer: “Only Peter, Paul, Jerry, and Terry were dealing drugs?”6 Id. at 35 (emphasis added). Once again defending the innocence of her son and nephew, she answered, “Yes.” That answer, given the context of the inquiry, hardly begins to establish Gaskins’s guilt. (The government extravagantly paraphrases this as testimony that Gaskins “was seen handling the crack many times”.) Moreover, Briscoe’s credibility is seriously open to question: she once denied to a grand jury that she saw drug transactions in her house, see id. at 34, 72, and, during the relevant period, she was on a number of medications (not to mention crack, see, e.g., id. at 17-18) under whose influence she had “a hard time remembering things sometimes”, id. at 38.
Reginald Chandler’s testimony was much more damning. He testified that although Gaskins was not a leading character in the conspiracy, he did perform a variety of ministerial, look-out, and enforcement functions. Yet Reginald’s testimony against Gaskins was somewhat vague; rather than narrating specific events and acts, he concentrated almost entirely on describing Gaskins’s general role within the conspiracy. See, e.g., 5/3 Tr. at 108-09, 118, 148-49; 5/4 Tr. at 58-60. But cf. 5/3 Tr. at 107. Anthony Chandler, by contrast, emphasized two specific moments in which he personally saw Gaskins perform his alleged enforcement role for the group. 5/4 Tr. at 88-90. The concrete nature of the testimony may have been decisive for the jury.
Quite apart from that concern, the jury might have been troubled by the fact that Reginald Chandler, who apparently could have been prosecuted with the others but wasn’t, agreed to testify after negotiating an agreement with the prosecution. See 5/3 Tr. at 92-94; 5/4 Tr. at 63. Along similar lines, the jury might have wondered why Reginald never even mentioned Gas-kins in earlier testimony before a grand jury. See 5/4 Tr. at 61-62. Because of these or other reasons, it might not have wished to convict Gaskins on the basis of what amounts to one person's arguably self-interested testimony.
To be sure, the jury apparently found Antonio guilty on the strength of the testimony of Briscoe and Reginald, who in turn attributed to Gaskins and Antonio similar roles in the conspiracy. One could argue from this that Antonio’s conviction embodies a kind of controlled experiment proving that the jury would have convicted Gaskins absent Anthony’s testimony. Yet the testimony against Gaskins was distinct from that against Antonio. Antonio, for instance, is said to have been part of the drug operation from the very beginning, see, e.g., 5/3 Tr. at 7, while Gaskins turned up later and, consequently, may have appeared less integral to the scheme. Anto
In short, Anthony Chandler’s testimony might have tipped the scales, and the jury’s exposure to his hallucinations might have produced a different outcome. We simply do not know. But as the issue is at least debatable, we will not deem the error harmless. We therefore reverse Gaskins’s conviction and remand for a new trial.
Appellants also argue that the trial court erred in instructing the jury on what it means, under 18 U.S.C. § 924(c), to use a firearm “in relation to” a drug trafficking crime. The court declared:
“In relation to” means that there must be a relation between the firearm and the underlying narcotic offense. You may infer that the possession of the firearm is intended to facilitate the underlying narcotic offense, but you are not required to do so.
5/8 Tr. at 75. Appellants raise two discrete objections to this instruction. First, they argue that the court should have required the jury, before convicting appellants on the § 924(c) count, to find that the relevant firearm was an “integral part” of the conspiracy’s drug trafficking crimes. Although several courts have used that metaphor in discussing § 924(c)(1), see, e.g., United States v. Munoz-Fabela, 896 F.2d 908, 911 (5th Cir.1990); United States v. Matra, 841 F.2d 837, 843 (8th Cir.1988), neither this circuit nor any other has required its use in the jury charge. Cf. United States v. Morrow, 923 F.2d 427 (6th Cir.) (finding charge substantially in the language of the statute to be plain error), vacated and ordered reheard en banc, 932 F.2d 1146 (6th Cir.1991).
Appellants’ more serious objection, which they raise in a single cryptic footnote, see Brief for Appellant Thomas at 11 n. 17, suggests that the second sentence of the instruction improperly permitted the jury to infer that any drug distributor who uses (or carries) a gun at all does so in connection with his drug operations. As the closest appellants came to raising the issue at trial, a vague assertion that they “would object vehemently to any inference given with respect to this charge”, 5/5 Tr. at 114, was too general to alert the trial court to their current claim, our inquiry is limited to determining whether the instruction amounted to “plain error”. See Fed.R. Crim.P. 52(b).
If the jury instruction clearly meant what appellants obscurely wish us to think it meant, it might have constituted reversible error. See United States v. Stewart, 779 F.2d 538, 540 (9th Cir.1985). Yet under a more probable interpretation, the instruction simply reminded the jury that, upon due deliberation, it had two choices: it could find the evidence enough to show the •necessary link between guns and drugs, or not enough. Read this way, the instruction arguably announced the obvious, but jury instructions often do that. By contrast, appellants’ suggested interpretation — that the instruction permitted the jury to choose between two contradictory legal standards — would render it absurd. At any rate, though not a model of clarity, the instruction did not constitute plain error.
Appellants’ remaining arguments do not merit discussion.
We affirm the convictions of appellants Thomas, Pryce, and Antonio, but reverse that of appellant Gaskins.
So ordered.
1.
Specifically, they were convicted of conspiring to distribute cocaine base and of conspiring to possess cocaine base with intent to distribute, in violation of 21 U.S.C. § 846; of possessing with intent to distribute 50 grams or more of cocaine base on December 20, 1988, in violation of 21 U.S.C. §§ 841(a)(1) and (b)(l)(A)(iii), and aiding and abetting the same, see 18 U.S.C. § 2; and of using and carrying a firearm during and in relation to a drug trafficking crime, in violation *86of 18 U.S.C. § 924(c), and aiding and abetting the same, see 18 U.S.C. § 2.
2.
At oral argument, the government lawyer effectively conceded as much, stating: "We’re not prepared in light of Davis v. Alaska, your honor, to say that the restrictions that were imposed would ultimately be upheld on appeal, because in all candor, I believe that Davis v. Alaska would not support these limitations.” Tr. of oral arg., 2/21/91.
3.
No one has claimed that counsel’s use of the report — the exact nature of which is unknown as a result of the trial court's ruling — would have violated any privilege of Anthony Chandler. Cf. D.C.Code § 14-307 (1981); Collins v. United States, 491 A.2d 480, 485 (D.C.App.1985). At any rate, the trial court restricted not only use of the report itself but also any reference to pre-December hallucinations.
4.
The role of the third factor is unclear, especially as one would expect that the prospect of "protracted, costly, and ultimately futile proceedings” would normally lead to some sort of consensual disposition, although the effect of the government’s paying the costs of both counsel (where that is the case) would reduce the likelihood.
5.
As an incidental matter, we note that appellant Thomas has never challenged the trial court’s' evidentiary ruling.
6.
These are the aliases of the four appellants. Gaskins is "Terry".