concurring in part and concurring in the result:
In my view, the Majority opinion’s ultimate determinations that the error in Roy’s case is subject to harmless-error review and that the error here was, in fact, harmless beyond a reasonable doubt are correct. But I respectfully disagree with the Majority opinion’s framework for evaluating cases raising claims involving counsel’s absence during the taking of directly inculpatory evidence in single-defendant trials.
Though the Majority opinion correctly acknowledges that the absence of counsel during the taking of directly inculpatory evidence can be structural error, see Maj. Op. at 1160-65, it rejects the notion that counsel’s absence can constitute the type of structural error known as Cronic1 error if counsel is not gone for the entirety of a “critical stage” of trial. Instead, the Majority asserts that we need a new substantial-portion-of-the-trial standard, derived without reference to Cronic, to assess whether structural error has occurred when counsel is absent for only part of trial. I cannot agree with this conclusion. If a defendant suffers deprivation of counsel that is “so likely to prejudice the accused that the cost of litigating [its] effect in a particular case is unjustified,” United States v. Cronic, 466 U.S. 648, 658, 104 S.Ct. 2039, 80 L.Ed.2d 657 (1984), then Cronic error has *1210occurred, regardless of whether the deprivation lasted for an entire “critical stage” of trial. As a result, we must presume prejudice.
The Majority’s development of a new standard to supplement Cronic solves a non-existent problem. Contrary to the Majority opinion’s contention, Cronic’s language does not impose a repressive “formula” that makes the opinion inapplicable in cases where counsel is absent for only part of trial. See Maj. Op. at 1160. The Majority opinion proceeds on the incorrect assumption that Cronic error occurs only when “defense counsel ‘entirely fails to subject the prosecution’s case to meaningful adversarial testing’ in the trial or where ‘the complete denial of counsel’ at a ‘critical stage of [the] trial’ ” happens. See id. at 1143-44 (quoting Cronic, 466 U.S. at 659, 104 S.Ct. 2089). But these formulations are merely demonstrative examples of “circumstances [involving denial of counsel] that are so likely to prejudice the accused that the cost of litigating their effect in a particular case is unjustified”— Cronic’s ultimate standard for structural error. Cronic, 466 U.S. at 658, 104 S.Ct. 2039.
In evaluating a deprivation-of-counsel error, we must not lose sight of our ultimate goal — to safeguard the adversarial process that gives a trial its basic character. The Supreme Court did not intend Cronic to provide an exhaustive list of specific circumstances giving rise to a presumption of prejudice. Rather, the decision and the Supreme Court’s later jurisprudence on structural error demonstrate that Cronic error includes any denial-of-counsel error that renders a trial presumptively unreliable because of a breakdown of the adversarial process. See id. at 656-58, 104 S.Ct. 2039. Simply put, when the absence is long enough to create a high probability that the accused fundamentally did not receive the trial promised to him under the Constitution, structural error occurs, and we need not conduct any search for actual prejudice. This type of error, of course, can occur when counsel is absent for only part of a critical stage of trial.
Nor are the differences between the Majority opinion’s approach and an analysis under Cronic merely semantic. The Majority opinion’s departure from Cronic imparts at least two undesirable consequences. First, the new standard that the Majority opinion announces today — the substantial-portion-of-the-trial standard— violates the Supreme Court’s instruction to use a categorical, rather than case-by-case, approach to determining whether an error is structural. Indeed, the Majority opinion’s test expressly requires case-by-case application and the weighing of subjectively judged factors. See Maj. Op. at 1164-65. This type of inquiry defeats the purpose of review for structural error — to identify and weed out circumstances highly likely to result in “fundamental unfairness” where finer-tooth review will often be impractical or judicially uneconomical. It will also necessarily cause inconsistent determinations about when structural error occurs in absent-counsel cases.
The Majority opinion’s substantial-portion-of-the-trial standard also does not sufficiently appreciate the fundamental nature of the absence-of-counsel error. So it relegates even lengthier absences to trial-error status, even though the role that counsel plays at trial warrants that all but the briefest of absences in a single-defendant trial constitute structural error.
Instead of the Majority opinion’s approach, we must evaluate whether counsel’s absence in a single-defendant trial justifies a presumption of prejudice without regard to whether the defendant was actually prejudiced in a given case. We do that by making a probability assessment of when, in general, counsel’s absence be*1211comes long enough that it is likely to result in a breakdown in the trial structure; create the appearance of unfairness to the jury and the public; affect counsel’s trial strategy; and make the potential for prejudice to the defendant high and the costs of litigating the actual effects of the denial, if even possible, not worthwhile. When that happens, the defendant has suffered structural error under Cronic.
Contrary to the Majority opinion’s suggestion, these factors allow hardly any wiggle room for the absence of counsel before trial error crosses the threshold of structural error and requires prejudice to be presumed. So structural error must be the rule, not the exception as the Majority opinion makes it, in absent-counsel cases.
I write separately to explain where the line must be drawn in the course of determining whether Cronic error has occurred and why it must be ascertained without respect to whether prejudice has, in fact, occurred in a given case. The general rule must be that counsel’s absence in a single-defendant trial is structural error under Cronic when it renders a trial presumptively unreliable because of a breakdown of the adversarial process — a rule that correlates with counsel’s absence for more than ten minutes or 1% of the trial.
1. Trial error and structural error differ in important ways.
Constitutional errors fall into two categories: trial error and structural error.2 Trial error happens “during the presentation of the case to the jury[ ] and ... may therefore be quantitatively assessed in the context of other evidence.” Arizona v. Fulminante, 499 U.S. 279, 307-08, 111 S.Ct. 1246, 113 L.Ed.2d 302 (1991). When trial error occurs, we evaluate it by determining whether the government has proven that the error was harmless to the outcome, beyond a reasonable doubt. Id. “[M]ost constitutional errors” fall into the category of trial error. United States v. Gonzalez-Lopez, 548 U.S. 140, 148, 126 S.Ct. 2557, 165 L.Ed.2d 409 (2006) (internal quotation marks omitted).
Structural error, in contrast, “affect[s] the framework within which the trial proceeds, rather than [being] simply an error in the trial process itself.” Fulminante, 499 U.S. at 310, 111 S.Ct. 1246. This type of error “necessarily render[s] a trial fundamentally unfair.” Rose v. Clark, 478 U.S. 570, 577, 106 S.Ct. 3101, 92 L.Ed.2d 460 (1986). For this reason, when structural error occurs, we do not give the government a chance to try to demonstrate beyond a reasonable doubt that the defendant was not prejudiced; instead, we assume prejudice without actually assessing the record for it. See generally Gonzalez-Lopez, 548 U.S. at 148, 126 S.Ct. 2557.
Failure to provide the “basic protections” at trial, id. — an impartial judge, the correct standard of proof, an impartial jury, and the assistance of counsel, see Neder v. United States, 527 U.S. 1, 9, 119 S.Ct. 1827, 144 L.Ed.2d 35 (1999)—results in structural error because when any of these safeguards is missing, “a criminal *1212trial cannot reliably serve its function as a vehicle for determination of guilt or innocence, and no criminal punishment may be regarded as fundamentally fair.” Rose, 478 U.S. at 577-78, 106 S.Ct. 3101 (citation omitted).
The effects of the structural error can be difficult or even impossible to quantify. Gonzalez-Lopez, 548 U.S. at 149 n.4, 126 S.Ct. 2557. And even when they can be assessed, structural error involves “ ‘circumstances ... that are so likely to prejudice the accused that the cost of litigating their effect in a particular case is unjustified.’ ” See Wright v. Van Patten, 552 U.S. 120, 124, 128 S.Ct. 743, 169 L.Ed.2d 583 (2008) (quoting Cronic, 466 U.S. at 658, 104 S.Ct. 2039).
Trial error and structural error differ in another important way as well. While we assess harmless error on a case-by-case basis, an error that qualifies as structural error does so categorically. As the Supreme Court has explained, “a constitutional error is either structural or it is not,” so we do not evaluate the specific impact of a given iteration of constitutional error upon a jury’s verdict when we determine whether the error constitutes structural error. Neder, 527 U.S. at 14, 119 S.Ct. 1827.
II. Denial of counsel during some of the taking of inculpatory evidence in a one-defendant trial can rise to the level of Cronic error.
Structural error can arise in different ways in the context of the denial of the Sixth Amendment right to counsel, but the touchstone for the analysis in all cases asks whether the denial has resulted in circumstances “so likely to prejudice the accused that the cost of litigating [its] effect in a particular case is unjustified.” Cronic, 466 U.S. at 658, 104 S.Ct. 2039. In Cronic, the Supreme Court identified some specific circumstances that would meet this standard to demonstrate how courts should think about the problem.
The “[m]ost obvious” variety of Cronic error arises when “the accused is denied counsel at a critical stage of ... trial.” Cronic, 466 U.S. at 659, 104 S.Ct. 2039. Perhaps this manifestation of Cronic error can occur when counsel is absent for a non-de minimis part of a “critical stage” of trial.3 But even if it cannot, counsel’s absence during a non-de minimis part of trial causes a “breakdown of the adversarial process,” which constitutes “circumstances that are so likely to prejudice the accused that the cost of litigating their effect in a particular case is unjustified.” Cronic, 466 U.S. at 657-58, 104 S.Ct. 2039. I explain each of these manifestations of Cronic error, in turn, below.
A. The denial of counsel during only part of a “critical stage” can rise to the level of Cronic error.
1. A “critical stage” is a discrete and readily identifiable, critically important unit of trial.
I agree with the Majority that a “critical stage” cannot consist of a single question and answer or even several questions and answers from a single witness. Maj. Op. at 1146. Rather, for the reasons the Majority opinion describes, a “critical stage” must be a discrete and readily discernible part of the trial. See id. at 1146-47.
*1213A “critical stage” of trial also “h[olds] significant consequences for the accused.” Bell v. Cone, 535 U.S. 685, 696, 122 S.Ct. 1843, 152 L.Ed.2d 914 (2002). For example, in discussing Cronic error that arises when “the accused is denied counsel at a critical stage of ... trial,” the Supreme Court has identified as “critical stages,” among others, arraignment, Hamilton v. Alabama, 368 U.S. 52, 54, 82 S.Ct. 157, 7 L.Ed.2d 114 (1961), the preliminary hearing, White v. Maryland, 373 U.S. 59, 60, 83 S.Ct. 1050, 10 L.Ed.2d 193 (1963), closing argument, Herring v. New York, 422 U.S. 853, 865, 95 S.Ct. 2550, 45 L.Ed.2d 593 (1975), and recess, Geders v. United States, 425 U.S. 80, 91, 96 S.Ct. 1330, 47 L.Ed.2d 592 (1976).
Here, the Majority opinion does not dispute that Roy was denied counsel while the trial court admitted directly inculpato-ry evidence in his counsel’s absence. Nor does the Majority opinion appear to contend that the taking of directly inculpatory evidence is not, as a whole, a “critical stage” of trial within the meaning of Cronic. And it could not.
As Judge Wilson points out, it is hard to envision a stage of trial that holds more “significant consequences” for the defendant, Bell, 535 U.S. at 696, 122 S.Ct. 1843, than the taking of directly inculpatory evidence. See Wilson Op. at 1237-39; see also Perry v. Leeke, 488 U.S. 272, 287, 109 S.Ct. 594, 102 L.Ed.2d 624 (1989) (Marshall, J., dissenting) (“ ‘[I]t is difficult to perceive a more critical stage ... than the taking of evidence on the defendant’s guilt.’ ”) (quoting Green v. Arn, 809 F.2d 1257, 1263 (6th Cir. 1987)). In the absence of inculpatory evidence, conviction is a legal impossibility because the defendant is presumed innocent until a jury finds that the government has presented sufficient evidence to establish guilt beyond a reasonable doubt. That cannot be said of other stages of trial, such as closing argument. And though a jury can return a guilty verdict without hearing closing argument by the prosecution, we have nonetheless held that closing arguments constitute a critical stage of trial. See Hunter v. Moore, 304 F.3d 1066, 1069-70 (11th Cir. 2002).
2. The denial of counsel during only part of a “critical stage” can rise to the level of Cronic error.
Though the Majority opinion does not dispute that the taking of inculpatory evidence meets the definition of a “critical stage” of trial, it asserts that the Supreme Court’s decision in Cronic “limited the presumption of prejudice to cases where defense counsel” was absent “during an entire ‘stage of [the] trial.’ ” Maj. Op. at 1143-44, 1146, 1147-48 (emphasis added) (quoting Cronic, 466 U.S. at 659, 104 S.Ct. 2039). But while the Majority opinion rejects the idea that counsel’s absence for less than an entire critical stage can constitute Cronic error, the Majority opinion nonetheless acknowledges that structural error can occur in those circumstances. See id. at 1147-48, 1160-61.
The Supreme Court has never held that the absence of counsel for part, but not all, of a critical stage of trial does not constitute structural error. First of all, it is not even clear that Cronic’s language supports the Majority opinion’s reading of Cronic to so limit the presumption of prejudice. The opinion speaks of the denial of counsel “at a critical stage of ... trial,” 466 U.S. at 659, 104 S.Ct. 2039 (emphasis added), not “throughout” a “critical stage.” And the denial of counsel for part of a critical stage is nonetheless the denial of counsel “at a critical stage.”
But more significantly, this language offers but one angle from which a court can approach the problem of defining structural error in denial-of-counsel cases. In no *1214case where the Supreme Court has found structural error concerning the right to counsel has the Court held that the absence of counsel for the entirety of a critical stage is a necessary prerequisite for a finding of structural error. ■ Instead, the cases to which the Majority opinion refers, and some of the Supreme Court’s descriptions of them, reflect only that those cases happened to concern facts involving the absence of counsel throughout the entire critical stage at issue.4
Although the Supreme Court has never considered a case like Roy’s, the Supreme Court has, at least once, effectively found structural error where the defendant suffered a deprivation of counsel for less than the entirety of what appears to be a “critical stage” of the proceedings. In Geders v. United States, 425 U.S. 80, 96 S.Ct. 1330, 47 L.Ed.2d 592 (1976), the defendant.was in the middle of his trial testimony when, despite defense counsel’s objections, the court prohibited the defendant from conferring with his counsel during a seventeen-hour overnight recess that occurred in the ten-day trial. Id. at 88, 96 S.Ct. 1330.
Without considering specific prejudice in Geders’s case in any way, the Supreme Court reversed the defendant’s conviction because the order “impinged upon [the defendant’s] right to the assistance of counsel guaranteed by the Sixth Amendment.” Id. at 91, 96 S.Ct. 1330. As the Court explained,
recesses are often times of intensive work, with tactical decisions to be made and strategies to be reviewed. The lawyer may need to obtain from his client information made relevant by the day’s testimony, or he may need to pursue inquiry along lines not fully explored earlier. At the very least, the overnight .recess during trial gives the defendant a chance to discuss with counsel the significance of the day’s events.
Id. at 88, 96 S.Ct. 1330. So Geders demonstrates that Cronic error can occur when a deprivation of counsel lasts for only part of a “critical stage.”
The Majority opinion attempts to distinguish Geders in two ways. Neither is persuasive.
First, the Majority notes that the opinion never used the terms “critical stage” or “stage” in its analysis. Maj. Op. at 1148. That’s true. Cronic had not been decided at that time, so Geders mentions neither “critical stages” nor “structural error.” But the Supreme Court has since indicated that it views Geders as part of its structural-error — and particularly the Cronic variety of its structural-error — jurisprudence.
Indeed, in Cronic itself, the Supreme Court specifically described its reasoning in Geders as having “found constitutional error without any showing of prejudice when counsel was either totally absent, or prevented from assisting the accused during a critical stage of the proceeding.” Cronic, 466 U.S. at 659 n.25, 104 S.Ct. 2039 (emphasis added). The fact that the Court did not use the terms “critical stage” or “stage” in Geders itself therefore does not somehow make Geders any less of an example of a deprivation of counsel that is presumptively prejudicial despite lasting through only part of a critical stage.
The Court’s ruling in Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984), reinforces this point about the relationship between Geders and Cronic. Strickland incorporates by refer*1215ence Cronic’s citation to Geders when it says, “Actual or constructive denial of the assistance of counsel altogether is legally presumed to result in prejudice.” 466 U.S. at 692, 104 S.Ct. 2052 (citing Cronic, 466 U.S. at 659 n.25, 104 S.Ct. 2039 (citing Geders, 425 U.S. at 80, 96 S.Ct. 1330)). The Majority opinion cites this sentence from Strickland and emphasizes the word “altogether” to support its theory that the Cronic exception applies only when counsel is absent for an entire “critical stage.” See Maj. Op. at 1145-46.
But Geders plainly does not fit that bill. In light of this fact and the specific issue that Strickland considered (ineffectiveness of counsel, not counsel’s actual absence for any length of time), the quoted Strickland sentence does not support the proposition that any presumptively prejudicial denial of counsel must last throughout an entire critical stage to be structural error. Rather, the Court in Strickland merely contrasted absent-counsel cases under Cronic with the situation where counsel is present and functioning throughout the trial but may be ineffective in some way.
Second, the Majority opinion characterizes Geders as “one of a line of decisions presuming prejudice where a defense attorney was prevented from, or impeded in, rendering assistance of counsel to his client because of an unconstitutional statute or court order.” Id. at 1148 (citing Perry, 488 U.S. at 279-80, 109 S.Ct. 594). This description is accurate as far as it goes, but it does not justify the Majority opinion’s subsequent conclusion that Ged-ers is simply a government-impediment case. Nor does it support the Majority opinion’s position that “[t]he ... statutory or court-ordered interference exception to the prejudice requirement that was applied in Geders ... does not apply in this case,” id. at 1149, even setting aside for the moment the fact that Cronic itself cites Geders in support of its critical-stage analysis. See Cronic, 466 U.S. at 659 n.25, 104 S.Ct. 2039.
Rather, Geders demonstrates that government impediment will nearly always, if not always, occur in some form in cases involving Cronic error. Among other descriptions, the Supreme Court has summarized Geders as a case “where [it] found a Sixth Amendment error without requiring a showing of prejudice” because Geders involved a criminal defendant “who had actually ... been denied counsel by government action.” Bell, 535 U.S. at 696 n.3, 122 S.Ct. 1843.
That also happened in Roy’s case. Starting trial and taking directly inculpatory evidence when the defendant’s counsel is absent — even inadvertently — likewise deprives a defendant of assistance of counsel through “government action.” It makes no difference whether the court,5 the prosecution,6 or defense counsel bears blame for *1216counsel’s absence.7 “Our Constitution places in the hands of the trial judge the responsibility for safeguarding the integrity of the jury trial,” United States v. Gainey, 380 U.S. 63, 68, 85 S.Ct. 754, 13 L.Ed.2d 658 (1965) (emphasis added), which includes “the duty of seeing that the trial is conducted with solicitude for the essential rights of the accused,” Glasser v. United States, 315 U.S. 60, 71, 62 S.Ct. 457, 86 L.Ed. 680 (1942), superseded on other grounds by statute as recognized in Bourjaily v. United States, 483 U.S. 171, 181, 107 S.Ct. 2775, 97 L.Ed.2d 144 (1987). Necessarily, then, “[t]he trial court should protect the right of an accused to have the assistance of counsel.” Id.
This makes perfect sense: the court alone enjoys control over the trial proceedings, including when to start, stop, and resume trial. And trial simply cannot proceed without the court’s actions in allowing it to do so. After all, the government, or even the defense, cannot call a witness to the stand and begin questioning when the trial judge is not present. Because of the judge’s essential role in convening trial, when a court conducts trial in defense counsel’s absence, government action has necessarily deprived a defendant of his right to counsel during trial. Inevitably, then, government impediment likely will always be relevant in absent-counsel cases.
But government impediment alone is not enough to explain why the error in Geders was structural. If it were, the error in Roy’s case would be structural without regard to the length of his counsel’s absence from trial. Indeed, the Majority probably would agree we should not conclude, based on Geders, that any government interference with the right to counsel always triggers structural error under any circumstances for essentially the same reasons the Majority opinion concludes that counsel’s absence for less than a “substantial portion of the trial” is not structural. See Strickland, 466 U.S. at 686, 104 S.Ct. 2052 (noting that “[gjovernment violates the right to effective assistance when it interferes in certain ways,” but not necessarily all ways, “with the ability of counsel to make independent decisions about how to conduct the defense”) (emphasis added). Consequently, Geders supports the notion that structural error — including Cronic error — can occur when an error lasts for only part of a critical stage.
In sum, the absence of counsel during part of a “critical stage” can constitute Cronic error.
*1217B. “[A]n actual breakdown of the adversarial process” results in Cronic error because the breakdoum amounts to “circumstances so likely to prejudice the accused that the cost of litigating their effect in a particular case is unjustified. ”
The question remains: how long of an absence during the taking of inculpatory evidence can be tolerated before the absence results in “circumstances so likely to prejudice the accused that the cost of litigating their effect in a particular case is unjustified”?
Unfortunately, as far as I can tell, no magical formula can tell us in every given case precisely where the breaking point is. But that does not mean we cannot meaningfully draw a probability line. After all, when we speak of structural error, we are talking about probabilities, not certainties. We must therefore conduct a probability assessment without respect to the particular facts of a given case because errors qualify as structural categorically. See Neder, 527 U.S. at 14, 119 S.Ct. 1827.
1. In the most basic and literal way, the denial of counsel in a single-defendant case during the taking of directly in-culpatory evidence undermines the adversarial process itself.
To conduct our probability assessment, we begin by reviewing why the Supreme Court delineated a category of structural errors in the first place: to make certain that the constitutional framework of procedural protections necessary for a fair trial remains intact. The Court was careful to note this broader goal in Cronic itself, in the specific context of the right to counsel: “the right to the effective assistance of counsel is recognized not for its own sake, but because of the effect it has on the ability of the accused to receive a fair trial.” 466 U.S. at 658, 104 S.Ct. 2039. So a presumptively prejudicial error is one that is highly likely to have “some effect ... on the reliability of the trial process.” Id.
In other words, it is one that is highly likely to “affect[] the framework within which the trial proceeds.” Fulminante, 499 U.S. at 310, 111 S.Ct. 1246. When we speak of this “framework,” we refer at a minimum to the anatomical features of the basic trial the Constitution envisions, unless a defendant chooses otherwise: an impartial jury, properly instructed on the prosecution’s burden of proving the defendant guilty beyond a reasonable doubt, see Sullivan v. Louisiana, 508 U.S. 275, 113 S.Ct. 2078, 124 L.Ed.2d 182 (1993); an impartial judge, see Rose, 478 U.S. at 577, 106 S.Ct. 3101 (citing Tumey v. Ohio, 273 U.S. 510, 47 S.Ct. 437, 71 L.Ed. 749 (1927)); and, of course, the “Assistance of Counsel for ... defence,” U.S. Const. amend. VI; see Cronic, 466 U.S. at 653 n.7, 104 S.Ct. 2039 (1984) (citing Gideon v. Wainwright, 372 U.S. 335, 344, 83 S.Ct. 792, 9 L.Ed.2d 799 (1963)).
The significance of Cronic — and the defining feature of “Cronic error” — lies in identifying a particular way in which denial of counsel so seriously affects the defendant’s ability to receive a fair trial that prejudice must be presumed. Under Cronic, a denial of counsel requires the presumption of prejudice when “an actual breakdown of the adversarial process” occurs during trial. Id. at 657-58, 104 S.Ct. 2039. “[I]f the process loses its character as a confrontation between adversaries,” id. at 656-57, 104 S.Ct. 2039, then the framework for trial envisioned by the Constitution collapses. See Fulminante, 499 U.S. at 310, 111 S.Ct. 1246; see also Framework, The American Heritage Dictionary of English Language (4th ed. 2000) (defining “framework” as a “structure for supporting ... something else”).
So our line must account for the crucial role that counsel plays in our trial frame*1218work. At no time is this role more important than during the taking of inculpatory evidence. Indeed, the taking of inculpatory evidence is perhaps the most critical part of the trial. Wilson Op. at 1238-39. Only evidence can convict an accused. So any tolerable absence cannot be too great before a trial loses its structural integrity as a “trial” under our Constitution. See Cronic, 466 U.S. at 656-59, 104 S.Ct. 2039.
When counsel is absent for any non-de minimis period during the taking of evidence, it is also, no doubt, obvious to the jurors and any spectating members of the public as well. This problem likewise causes dangerous cracks in our trial’s foundation because it conflicts with the court’s “independent interest in ensuring that criminal trials are conducted within the ethical standards of the profession and that legal proceedings appear fair to all who observe them.” Gonzalez-Lopez, 548 U.S. at 152, 126 S.Ct. 2557 (emphasis added) (internal quotation marks omitted); see Wilson Op. at 1247 (quoting Indiana v. Edwards, 554 U.S. 164, 177, 128 S.Ct. 2379, 171 L.Ed.2d 345 (2008) (citation omitted)).
A lengthy absence of counsel may, as Judge Wilson points out, see id. at 1235-36, cause the jury to develop its own ideas about the significance of defense counsel’s absence from trial — that defense counsel may not believe in the defendant’s case, that the court thinks so little of the defendant or his counsel that it does not deem it worthwhile to wait for counsel before beginning, that the case itself is unimportant and not worthy of the formality otherwise attached to criminal trials, or that any number of other unfair ideas justify resuming trial in the absence of defense counsel. While I do not suggest that juries do not follow instructions to consider only the admitted evidence, that does not mean that factors such as these have no subconscious effect on their thinking. Indeed, I see no reason why these unfair prejudices would be any less threatening to the rights of the accused than the ones that the Federal Rules of Evidence explicitly seek to avoid. See, e.g., Fed. R. Evid. 403.
Nor does the fact that a record of what happened while our judicial structure caved in on itself during the absence of counsel somehow remedy this framework problem. See Maj. Op. at 1161-62 (suggesting that the fact that what counsel missed in his absence can be determined “should bear heavily on whether to presume prejudice”). So in order to account for the trial-framework problem, any absence must be brief to avoid rising to the level of Cronic error.
Yet Cronic’s use of the phrase “an actual breakdown of the adversarial process” contemplates more than the momentary unavailability of counsel. As Cronic explains, “The right to the effective assistance ’ of counsel is ... the right of the accused to require the prosecution’s case to survive the crucible of meaningful adversarial testing.” 466 U.S. at 656, 104 S.Ct. 2039. And “meaningful adversarial testing” can occur at trial even if counsel is absent for a brief period.
Indeed, some absences may be so short that they cannot fairly be viewed by any measure as affecting the framework within which the trial proceeds because defense counsel is present to provide assistance throughout literally nearly all of trial. To take an extreme example, if counsel misses five seconds of testimony, the trial structure itself does not collapse. Rather, the trial maintains its character as an adversarial proceeding. Little occurs in counsel’s absence, and we can easily evaluate the significance of the five seconds’ worth of testimony taken. A jury is similarly unlikely to draw negative inferences from such an absence. Presuming prejudicé under these circumstances makes little sense. *1219The same is true of slightly longer absences, up to a few minutes. For the reasons I have discussed, though, soon after that, counsel’s absence necessarily begins to cause “an actual breakdown of the adversarial process,” since our trial framework can withstand one of its structural pillars missing for only so long before it starts to crumble irreparably. See Cronic, 466 U.S. at 657-58, 104 S.Ct. 2039.
2. Evaluating the effects of counsel’s absence during the taking of inculpatory evidence at a single-defendant trial becomes challenging or impossible the longer the absence persists.
Second, we must draw our line at a point before the absence has lasted long enough to interfere with our ability to assess its effects. See Gonzalez-Lopez, 548 U.S. at 149 n.4, 126 S.Ct. 2557 (noting that “the difficulty of assessing the effect of the error” is a basis for identifying structural error). As with the trial framework’s tolerance of counsel’s absence, that point arrives soon after the absence begins.
Although the Majority opinion concludes that the harm resulting from counsel’s absence is limited to the erroneous admission of evidence, see Maj. Op. at 1157-58 (quoting Satterwhite v. Texas, 486 U.S. 249, 257-58, 108 S.Ct. 1792, 100 L.Ed.2d 284 (1988)), that conclusion does not recognize either the harm to the trial framework that I have discussed above or the harm to the full scope of counsel’s representation that can occur when more than a brief absence happens.8 Judge Wilson eloquent*1220ly makes these points in his Dissent. See Wilson Op. at 1236-37.
As Judge Wilson notes, see id., counsel’s responsibility during trial does not consist solely of keeping out objectionable evidence. Defense counsel orchestrates the entire defense, of which challenging objectionable evidence is but a single part. Among other things, in the heat of trial, defense counsel must make necessary adjustments to the defense strategy in real time, in light of the happenings in court; tailor cross-examination of witnesses, in part, to the witnesses’ testimony on direct examination during trial; evaluate on an ongoing basis the advisability of putting on and the contents of any defense case, including presenting the defendant to testify on his own behalf; determine whether to address and, if so, how to account during closing argument for evidence admitted during trial; respond to questions and concerns of his client; clear up any misunderstandings his client may have; and detect and respond to cues from the jury’s demeanor. Cf., e.g., Van Patten, 552 U.S. at 127 n.*, 128 S.Ct. 743 (Stevens, J., concurring). When counsel is absent, no one is doing any of these things, and, if the absence extends for much more than a few minutes, all of these functions can be impeded even upon counsel’s return.
The Majority opinion’s conclusion that the harm incurred during counsel’s non-de minimis absence is limited to the erroneous admission of evidence also ignores the realities of trial. Trial is not like a brick wall, made up of many fungible parts that can be easily interchanged and reordered with necessarily the same end result. Trial is a living, developing thing. See Geders, 425 U.S. at 86, 96 S.Ct. 1330 (“A criminal trial does not unfold like a play with actors following a script; there is no scenario and can be none[;] ... complexities and contingencies [are] inherent in the adversary process.”). What happens or does not happen at one point of a trial can deeply affect the proceedings that follow.
As a result, in a single-defendant trial, the harm from a non-de minimis absence of counsel is not confined to a simple erroneous admission of evidence at trial. Rather, the erroneous admission of evidence in counsel’s absence is but one manifestation of the harms counsel’s absence inflicts in such circumstances, much like a cough is often but one symptom of tuberculosis. Considering only the effect of erroneously admitted evidence during counsel’s non-de minimis absence is a lot like treating a tuberculosis patient with nothing more than cough drops.
Nor does the Majority opinion’s observation that many of the errors that might result from counsel’s non-de minimis absence are themselves subject to harmless-error review (including lost objections, “hampered cross-examination,” and lost impeachment arguments) remedy the assessment problem. Maj. Op. at 1165-74. Rather, this error-by-error piecemeal analysis misses the forest for the trees: as counsel’s absence grows longer, we cannot know the precise brew of constitutional error that’s been allowed to ferment. The Majority opinion would have us hold the balance of error in equipoise and analyze each particular type of error on its own. But as the absence grows longer, there is no control for the other types of error that occurred because of counsel’s absence. We cannot assess the magnitude of a particular type of error in light of the rest of the evidence, as we must in harmless-error review, because in these circumstances, counsel’s absence injected an unknowable concoction of error into the trial.
*1221Notably, the speculation in which we must indulge when a non-de minimis absence occurs is not the type of guided speculation we engage in under, for example, Strickland,9 when we evaluate whether counsel’s ineffective choice may have prejudiced the defendant’s case. Instead, no record exists regarding what counsel did or did not do, and we have nothing to compare to the vast range of choices a reasonably competent attorney could have made. We must entirely imagine the many options available to competent counsel throughout the duration of counsel’s absence — a task that would be sure to miss some viable alternatives. Then we must use our imaginations to guess how each possible choice might have caused counsel to modify his approach to the rest of the defense at trial. This is pure speculation, three or four times removed from the circumstances contemplated in Strickland.
Then — and only then — do we arrive at the next part of the speculation: how counsel’s imagined Neverland10 performance may have affected the outcome of the trial. This is like trying to guess how going back in-time would affect the space-time continuum.11 But our Constitution does not abide *1222a world of imagined lawyers in alternative universes when it comes to a right so dear as that of “Assistance of Counsel for [an accused’s] defence.”
Rather, our system can tolerate only a brief absence before our ability to evaluate the effects of the absence enter this speculative realm. Nevertheless, some absences are so brief — a few seconds or even minutes — that the effects are not necessarily incapable of being evaluated. The types of prejudice that might occur under these circumstances — a particularly prejudicial line of questioning, the introduction of an especially prejudicial exhibit, an unfair characterization, etc. — are different in kind than the subtler, more insidious harms introduced by the longer absence of counsel' in a single-defendant trial and are therefore susceptible of harmless-error review. Counsel’s viable options for dealing with what occurred in his absence under such circumstances are likewise far more limited than once the absence extends much more than a few minutes. A very brief absence allows us to identify what counsel’s options might be upon her return without resorting to rank speculation. But the universe of options expands exponentially as the period of absence grows.
So while cutting out the localized cancer of a very brief absence is possible, the effects of an absence metastasize throughout the trial in ways that are no longer readily identifiable once the length of absence lasts longer than a few minutes. As a result, unlike with a very brief absence, the likelihood of prejudice from an absence that lasts more than a few minutes substantially increases, and the ability to identify the resulting prejudice markedly decreases. This factor likewise supports drawing a line that differentiates absences that are just a few minutes from those lasting longer.
3. Because the taking of directly inculpa-tory evidence in counsel’s absence in a single-defendant trial quickly becomes highly likely to result in prejudice, and detecting the absence of counsel in a single-defendant trial is extremely easy, the point where it is not worth litigating the effects of this category of error in a given case must come not long after counsel’s absence begins.
The miniscule costs associated with setting the threshold for structural error in absence-of-counsel cases shortly after counsel’s absence begins in a single-defendant trial also warrants drawing a line not long after counsel’s absence begins.
For the reasons I have already described, counsel’s absence in a single-defendant trial during the taking of directly inculpatory evidence will quickly introduce a significant and unquantifiable mix of prejudice into a single-defendant trial. Yet absent defense counsel during the taking of inculpatory evidence in a single-defendant trial is an error that is “easy to identify” and therefore “easy for the government [including the court and the prosecution] to prevent” — not coincidentally another hallmark of structural error. Strickland, 466 U.S. at 692, 104 S.Ct. 2052. The utter lack of any defense counsel at the start of trial proceedings should be immediately obvious to both the judge and the prosecution.
*1223First, we are not speaking of some trivial technical requirement. We are talking about a fundamental constitutional right that should be — and no doubt is — always at the tops of the minds of the trial judge and the prosecution during trial: the right to counsel. Indeed, as I have noted, the court has an affirmative obligation to protect a defendant’s right to counsel during trial. See supra at 1215-16 (citing Gainey, 380 U.S. at 68, 85 S.Ct. 754; Glasser, 315 U.S. at 71, 62 S.Ct. 457).
Second, we are not looking for a needle in a haystack. Visually, the absence of counsel is stunningly obvious. Detecting the absence of sole counsel in a single-defendant trial is as straightforward as looking at the defense table. When no defense counsel is present in the courtroom, only one person sits at the defense table— the defendant — and counsel’s absence is conspicuous.
Third, judges can and often do ask counsel for both parties whether they are ready to proceed before bringing in the jury. Even if a judge and prosecutor do not notice the absence of counsel before the inquiry, the lack of a response from defense counsel at that time would certainly alert them to counsel’s absence.
But perhaps the greatest indication that the error is “easy to identify” consists of the fact that neither the Majority opinion nor the parties are able to cite a single case other than Roy’s where directly incul-patory evidence was taken in the absence of sole counsel in a one-defendant trial. Simply put, this error is so obvious that it quite literally almost never happens.12 And because district courts and the prosecution are so well attuned to the need for counsel’s presence in a single-defendant trial that they are highly likely to notice counsel’s absence immediately or, at worst, very shortly after trial resumes, drawing the structural-error line not long after counsel’s absence begins in a single-defendant trial imposes virtually no costs.
4. Consideration of all of the factors that cause counsel’s absence during a single-defendant trial to merit a presumption of prejudice supports drawing the line between trial error and structural error at counsel’s absence that lasts for more than ten minutes or 1% of the combined “critical stages” of trial.
All of the factors in determining when a presumption of prejudice is appropriate in absent-counsel cases have at least one thing in common: they all point to a very low threshold of tolerance for absence of counsel during a “critical stage” of a one-defendant trial before the absence crosses the line from trial error to structural error. But that threshold is not zero. Rather, for the reasons I have explained, harmless-error analysis can effectively and appropriately be performed when counsel’s absence lasts only a few minutes.
So the probability line where an absence becomes long enough to create “circumstances that are so likely to prejudice the accused that the cost of litigating their effect in a particular case is unjustified” must be drawn soon after counsel’s absence spans more than a few minutes. And since this line must be ascertainable in any case without requiring any type of actual prejudice review of the record, I would draw the line when counsel is gone in a single-defendant trial at the lesser of either more than ten minutes or more than 1% of the combined critical-stage portions *1224of trial. When ten minutes constitutes 1% or less of the combined critical stages of trial, the period is brief enough that the admitted evidence is relatively little, counsel can quickly and easily learn what he has missed and adjust his strategy accordingly, and the appearance of fairness and integrity in the trial is not undermined. A jury may reasonably infer, for example, that counsel has simply stepped out to use the restroom or check on a witness.
Though the period where an absence truly becomes structural error is surely greater than ten minutes or 1% of the taking of evidence, the probability of diminishing returns from attempting to conduct a prejudice analysis begins to increase significantly not long after counsel’s absence lasts for ten minutes or 1% of the trial. And since we cannot identify a strict cutoff that necessarily includes only trial errors on one side and only structural errors on the other, we must err on the side of including some trial-error absences in the structural-error category, rather than the other way around. After all, we are discussing a constitutional violation that is serious enough and the effects of which are difficult enough to assess that where it is found, prejudice is presumed. And, significantly, it is an error that is easily preventable, so the costs of setting a low threshold are negligible.
When we apply this line to Roy’s case, we find that his case involves an absence that does not cause a breakdown of the adversarial process or any other damage to the fundamentally fair character of his trial. The counsel’s absence in Roy’s case did not create “circumstances that are so likely to prejudice the accused that the cost of litigating their effect in a particular case is unjustified.”
Roy’s counsel was absent for a total of seven minutes, so he was missing for less than ten minutes of the total of all critical stages of trial. And as a percentage of the total critical stages of trial, those seven minutes amounted to less than 1%. Counsel’s absence therefore falls on the trial-error side of the error line.13 As a result, we conduct harmless-error review in Roy’s case.
Nevertheless, the mere fact that a case may be susceptible of harmless-error review does not mean, of course, that any error is necessarily harmless. To the contrary, where harmless-error review applies, the court must be convinced that “on the whole record ... the error ... [is] harmless beyond a reasonable doubt.” Rose, 478 U.S. at 583, 106 S.Ct. 3101 (citation and quotation marks omitted). If the court cannot satisfy itself in this way— either because the record suggests that the error was not harmless beyond a reasonable doubt or because the record as a whole does not provide sufficient information to allow a determination to be made— even a trial-error absence will require reversal and remand for a new trial.
But that is not the case here. As the Majority ably explains, the record here *1225clearly demonstrates that Roy’s counsel’s trial-error absence was harmless beyond a reasonable doubt.
III. The fact that a brief absence of counsel during part of a “critical stage” of a single-defendant trial quickly rises to the level of Cronic error does not necessarily mean that the same thing is true in a multi-defendant trial.
The Majority opinion worries that recognizing that the structural-error threshold is low for counsel’s absence during trial in a single-defendant trial means that the threshold must be set equivalently low in multi-defendant trials. This question is not before us, so I do not offer an opinion on it. Nevertheless, I express some thoughts as to why I do not share the Majority opinion’s concerns, homing in on the factors that determine whether an absence of counsel during a critical stage of trial is structural error in the first place.
Beginning with the basic trial framework, when one or more defense lawyers are present during a multi-defendant trial, the overall trial structure itself at least arguably remains intact, even if other defense counsel are absent. That is, some licensed attorney serves in an adversarial role against the prosecution and might, as a practical matter, simultaneously assist in the defense of other defendants while acting on behalf of her own client.14 But when a single defendant has no counsel whatsoever where directly inculpatory evidence is offered, courtroom proceedings do not even look like a “trial” as our Constitution envisions it, and counsel’s absence very quickly rises to the level of Cronic error. No one is present to assist even theoretically in the accused’s defense.
Second, while the absence of a particular defendant’s attorney is still ascertainable when multiple defendants and defense counsel are present, one defendant’s sole missing defense attorney at a table of, for example, five defendants and seven counsel,15 is not as visually conspicuous as a sole defendant’s appearance all alone at the defense table during trial. As a result, the point where the proceedings cease to appear like a constitutional trial to the jury and public is certainly higher and, depending on the circumstances, possibly nonexistent.
Third, although the effect on the proceedings of a single defendant’s counsel’s longer absence in a multi-defendant trial may still be challenging and difficult to assess, at least we can review a record of how some defense counsel reacted to the questioning, the jury, and, where applicable, the client during the absence, so our speculation is not necessarily entirely imagined, and some form of a Strickland-type of analysis of the present attorney’s actions may perhaps be possible. Similarly, at least the defendant’s absent counsel can consult a professionally trained, defense-oriented person (a defense attorney who was present during the absence) about what transpired in his absence, so he can adjust his defense accordingly. These things are not even possibilities where counsel is absent for a non-de minimis *1226portion of the taking of inculpatory evidence in a single-defendant trial.
Fourth, multi-defendant trials are often significantly longer than single-defendant trials. If counsel for more than one defendant in a multi-defendant trial is absent for more than a brief period of the taking of directly inculpatory evidence, the cost and effort of attempting to evaluate the-record for harmless error may be justifiable in a way that it is not many single-defendant trials.
For these reasons, I respectfully disagree with the Majority opinion that recognizing Cronic error when counsel is briefly absent during part of a single-defendant trial dictates that counsel’s brief— or even longer — absence in a multi-defen-dant trial would then also necessarily qualify as Cronic error or some other type of structural error.
IV. The Majority opinion’s solution for determining when counsel’s absence during part of a “critical stage” constitutes structural error is flawed because it is not categorical and because it sets too high a threshold for structural error when counsel is denied during a single-defendant trial.
The Majority opinion holds that counsel’s absence during part of trial rises to the level of structural error when counsel misses a “substantial portion of the trial,” determined “on a case-by-case basis considering, among other factors, the length of time counsel was out, the proportion of the trial missed, and the significance of what he missed.” See Maj. Op. at 1165. It further suggests through its analysis that a “substantial portion of the trial” is a relatively long period. In my view, both of these conclusions are inconsistent with the Supreme Court’s jurisprudence on structural error involving the denial of counsel.
First, as I have mentioned, the Supreme Court has cautioned against “importing] into the initial structural-error determination (ie., whether an error is structural) a case-by-case approach that is more consistent with our traditional harmless-error inquiry (ie., whether an error is harmless). Under [the Supreme Court’s] cases, a constitutional error is either structural or it is not.” Neder, 527 U.S. at 14, 119 S.Ct. 1827 (characterizing the Supreme Court’s “traditional ... approach to structural errors” as “categorical”).
And this makes sense. If determining in the first place whether a type of error was structural or trial required an analysis of actual prejudice in a given case, it would not differ from harmless-error analysis: in any case where error was actually assessable but was, beyond a reasonable doubt, harmless to the defendant, the error would be harmless, and in any case where the prejudice inflicted by the error was either not assessable or demonstrably resulted in prejudice to the defendant, the error would be harmful.
But that’s not how the dichotomy between structural error and trial error works. Rather, structural-error jurisprudence recognizes fundamental errors and requires us to make a probability determination that the existence of that type of error in general creates “ ‘circumstances ... that are so likely to prejudice the accused that the cost of litigating their effect in a particular case is unjustified.’ ” Van Patten, 552 U.S. at 124, 128 S.Ct. 743 (quoting Cronic, 466 U.S. at 658, 104 S.Ct. 2039). Engaging in any attempt to calculate the actual prejudicial effects of a type of error to determine whether, in a given case, it qualifies as structural in nature defeats the purpose of categorizing particular types of errors as structural.
The Majority opinion’s solution to when structural error occurs, however, expressly *1227calls for “case-by-case ... consideration], [accounting for,] among other factors, the length of time counsel was out, the proportion of the trial missed, and the significance of what he missed.” Maj. Op. at 1165 (emphasis added). This approach necessarily requires the court to conduct some type of individualized assessment of prejudice in a given case to determine whether the error as presented in that case constitutes structural error, even though the Supreme Court has warned against delineating the parameters of a category of structural error by assessing the evidence adduced in a particular ease.
Besides this problem, the Majority opinion’s approach employs a balancing test, so it will necessarily yield conflicting results concerning whether an absence qualifies as structural error, depending on who applies the test, how the judge construes each factor, and how she or he weighs the test’s factors. For example, what length of absence is too long and how do we decide? What proportion of trial is too great? How do we judge the “significance of what [counsel] missed”? Is determining the “significance of what [counsel] missed” some form of a mini-harmless-error inquiry? How do we balance the four expressly named considerations against each other? What other factors should be considered, and how are they weighed in the balance?
And since knowing what counsel missed is “at least as important” a factor as the other three and “should bear heavily on whether to presume prejudice,” see id. at 1162, does it outweigh a longer absence that comprises a good percentage of the trial? Different judges applying the substantial-portion-'of-the-trial factors will, of course, arrive at different conclusions about whether structural error has occurred in any given case — a red flag that the line for structural error has not been categorically drawn.
Nor does the Majority opinion’s application of the substantial-portion-of-the-trial test to Roy’s facts provide much guidance. Instead, it simply observes that we know what counsel missed and reduces counsel’s absence to numbers:
Roy’s counsel missed only seven minutes of a trial that lasted 1,884 minutes or 31.4 hours (not counting recesses and jury deliberations), which is less than one-half of one percent of trial time. He missed only 18 answers that were given by one of the government’s 13 witnesses who collectively gave a total of approximately 2,745 answers, meaning he missed less than one percent of the total. And we know exactly which questions and answers he missed. His physical absence was far more momentary and far less substantial than any in the five cases that our sister circuits have decided under the substantial portion standard. We have no trouble concluding that Roy’s counsel did not miss a substantial portion of the trial.
Id. at 1165-66.
Other than the fact that the Majority opinion adjudged Roy’s counsel’s absence “far more momentary and far less substantial than any in the five cases that our sister circuits have decided under the substantial portion standard,” we don’t know how the Majority weighed the factors against each other; how “miss[ing] only 18 answers that were given by one of the government’s 13 witnesses who collectively gave a total of approximately 2,745 answers, meaning he missed less than one percent of the total,” tells us the “significance of what [counsel] missed” any more than the number of minutes missed and the percentage of trial missed; what other factors we should consider when conducting this analysis; or when the fact that a record of what counsel missed exists ceases to support harmless-error review. And *1228in the cavernous abyss between Roy’s 7-minute absence and the other circuits’ substantial-portion-of-the trial cases, where the attorney slept for either more than a day or slept repeatedly for several minutes at a time throughout the entire trial, we don’t know where a trial-error absence becomes a structural-error absence.
The Majority opinion’s substantial-portion-of-the-trial test also suffers from another problem: it significantly undervalues the right to counsel during trial, so it sets the bar too high for when counsel’s absence crosses the threshold from trial error to structural error. As I have previously explained, the right to counsel during trial is essential under our system of justice, and it does not take long for counsel’s absence from part of trial to create serious, exponentially multiplying problems “that are so likely to prejudice the accused that the cost of litigating their effect in a particular case is unjustified.” See supra at 1217-23; Cronic, 466 U.S. at 658, 104 S.Ct. 2039. For this reason, only a brief absence can be tolerated in a single-defendant trial before the likelihood of prejudice greatly outweighs the benefits of attempting to engage in a prejudice analysis.
The Majority opinion turns the significance of the right to counsel during a single-defendant trial upside down, essentially creating a rule under which counsel’s absence — even for long periods — constitutes nothing more than trial error, except in the most extreme circumstances. But the right to counsel — particularly during trial — is absolutely fundamental to our system of justice. A single-defendant trial where counsel is absent for more than a very brief period inflicts great damage upon our system of justice; it is antithetical to it, to our sense of fairness, and to the reliability of any resulting verdict. Does the fact that we know what happened when counsel was gone somehow negate the deleterious effects on the trial framework of a long absence that comprises a good percentage of the trial? I think not.
And because of the ease with which defense-counsel absences in a single-defendant trial can and should be prevented, even less justification exists for tolerating anything more than counsel’s very brief absence. Since the substantial-portion-of-the-trial test that the Majority opinion adopts today to determine whether an attorney’s absence qualifies as structural error does not sufficiently value the right to counsel during a single-defendant trial, I respectfully disagree with that standard.
V. Conclusion
So I end where I began. I concur in the Majority’s conclusion that the error in this case was harmless because it was not long enough to rise to the level of Cronic error, and the record shows it to have been harmless beyond a reasonable doubt.
But in a single-defendant trial, the non-de minimis absence of counsel creates “circumstances that are so likely to prejudice the accused that the cost of litigating their effect in a particular case is unjustified.” It collapses the constitutional framework of the trial; is easily identifiable and preventable by the court and the government; introduces an unknowable mix of error into the trial that is so likely to prejudice a defendant that assessing its effects in any given case is not worthwhile; and renders the trial process unreliable amounts to structural error under Cronic. Based on these considerations, I would draw the line between trial-error absences and structural-error absences at the point where an absence lasts for more than ten minutes or 1% of the total “critical stages” of trial. In my view, this approach comports with Cronic, the categorical nature of structural error as the Supreme Court has explained it, and the importance of the right to counsel during trial.
. United. States v. Cronic, 466 U.S. 648, 104 S.Ct. 2039, 80 L.Ed.2d 657 (1984).
. In United States v. Gonzalez-Lopez, 548 U.S. 140, 159, 126 S.Ct. 2557, 165 L.Ed.2d 409 (2006) (Alito, J., dissenting), four dissenting Justices took issue with the Court’s division of constitutional error into the all-inclusive categories of trial error, which always is subject to harmless-error review, and structural error, which always results in automatic reversal. While the Court stated that its conclusion that denial of the right to counsel of choice constitutes structural error '‘relie[d] neither upon such comprehensiveness nor upon trial error as the touchstone for the availability of harmless-error review,” the Court nonetheless responded to the dissenters’ objection by opining that “it is hard to read [its precedent] as doing anything other than dividing constitutional error into two comprehensive categories.” Id. at 149 n.4, 126 S.Ct. 2557.
. As I discuss later in this concurrence, Cronic speaks in terms of circumstances that result in an “actual breakdown of the adversarial process” and that are “so likely to prejudice the accused that the cost of litigating their effect in a particular case is unjustified.” 466 U.S. at 657-68, 104 S.Ct. 2039. The absence of counsel for a very brief period does not effect these consequences, so the denial of counsel during part of a critical stage of trial that Cronic speaks of must be more than de minimis.
. See, e.g., Holloway v. Arkansas, 435 U.S. 475, 98 S.Ct. 1173, 55 L.Ed.2d 426 (1978) (counsel labored under conflict of interest throughout entire proceeding); White, 373 U.S. at 59-60, 83 S.Ct. 1050 (counsel absent for an entire preliminary hearing); Hamilton, 368 U.S. at 52, 82 S.Ct. 157 (counsel absent for entire arraignment).
. Here, the court began trial a minute earlier than the scheduled recess ended. Counsel certainly should have been present at that point, and efficiency and promptness are praiseworthy qualities in a district court. Nevertheless, if blame were relevant — it's not for the reasons I have mentioned — a court that starts trial before the end of a scheduled recess, when counsel in a single-defendant trial is absent, is not entirely without fault in the deprivation.
. Nevertheless, "[t]he United States Attorney is the representative not of an ordinary party to a controversy, but of a sovereignty whose obligation to govern impartially is as compelling as its obligation to govern at all; and whose interest, therefore, in a criminal prosecution is not that it shall win a case, but that justice shall be done. As such, he is in a peculiar and very definite sense the servant of the law.” Berger v. United States, 295 U.S. 78, 88, 55 S.Ct. 629, 79 L.Ed. 1314 (1935). Consequently, the prosecutor has an independent responsibility to ensure that trial does not proceed when defense counsel is not present.
. That is not to say that counsel who is late may not be sanctioned. It is certainly fair and appropriate for a judge to expect attorneys to be on time. When an attorney is late, the resulting delay unnecessarily takes up the court’s precious time and unfairly impinges on the jury's time. Of course, sometimes circumstances beyond an attorney’s control can occur. Nothing in the record indicates that defense counsel in Roy's case was chronically late for trial or regularly returned late from recesses. If it did, sanctions of some type might well be in order. But even if they were, “[tjhere are other ways to deal with the problem ... short of' denying the defendant his counsel during the taking of directly inculpa-tory evidence. Cf. Geders, 425 U.S. at 89, 96 S.Ct. 1330. For example, the court could monetarily sanction counsel outside the presence of the jury, or, if counsel is court-appointed, the court could remove counsel from the approved Criminal Justice Act list. But counsel's tardiness alone cannot justify denial of a defendant’s right to counsel. Of course, if a defendant affirmatively schemes to introduce reversible error through his counsel’s absence, that is a different matter and may constitute a waiver of the right. The record contains no evidence that that is the situation here.
. The Majority opinion’s reliance on Fulmi-nante, 499 U.S. at 306-07, 111 S.Ct. 1246, and cases it cites to demonstrate that the erroneous admission of evidence is subject to harmless-error review is flawed for the same reasons. See Maj. Op. at 1167. The kinds of error at issue in Fulminante and the cases it cites tell us nothing about whether the absence of counsel during part of trial constitutes structural error. Not one of the opinions identified in Fulminante indicates that counsel was not present at the actual trial, when the challenged evidence against the defendant was admitted. So when the errors in those cases occurred, the structural framework of the trial was intact and counsel was aware of and able to confront the fallout from the erroneous admission of evidence, unlike when an absence of counsel occurs in a single-defendant trial during part of the taking of directly inculpatory evidence. The Majority opinion’s reliance on Florida v. Nixon, 543 U.S. 175, 125 S.Ct. 551, 160 L.Ed.2d 565 (2004), Bell, 535 U.S. at 688, 122 S.Ct. 1843, and Cronic, 466 U.S. 648, 104 S.Ct. 2039, 80 L.Ed.2d 657, suffers from a similar problem: counsel was present at trial when the challenged actions occurred. In fact, it was counsel's actions during his presence at trial that were at issue in those cases. Similarly, counsel was present at trial when the errors happened in all of the other cases the Majority opinion cites in support of its position that harmless-error analysis applies to the absence-of-counsel error in all but those cases where counsel was absent for a substantial portion of the trial. See Maj. Op. at 1169-74 (citing Hinton v. Alabama,U.S. -, 134 S.Ct. 1081, 188 L.Ed.2d 1 (2014) (per curiam); Harrington v. Richter, 562 U.S. 86, 131 S.Ct. 770, 178 L.Ed.2d 624 (2011); Banks v. Dretke, 540 U.S. 668, 124 S.Ct. 1256, 157 L.Ed.2d 1166 (2004); Strickler v. Greene, 527 U.S. 263, 119 S.Ct. 1936, 144 L.Ed.2d 286 (1999); United States v. Bagley, 473 U.S. 667, 105 S.Ct. 3375, 87 L.Ed.2d 481 (1985); Jones v. Butler, 778 F.3d 575 (7th Cir. 2015); Barwick v. Secy, Fla. Dep’t of Corr., 794 F.3d 1239 (11th Cir. 2015); United States v. Travillion, 759 F.3d 281 (3d Cir. 2014); Gissendaner v. Seaboldt, 735 F.3d 1311 (11th Cir. 2013); Roberts v. Comm’r, Ala. Dep’t of Corr., 677 F.3d 1086 (11th Cir. 2012); Ponticelli v. Sec’y, Fla. Dep’t of Corr., 690 F.3d 1271 (11th Cir. 2012); Boyd v. Comm’r, Ala. Dep’t of Corr., 697 F.3d 1320 (11th Cir. 2012); United States v. Orr, 636 F.3d 944 (8th Cir. 2011); Pietri v. Fla. Dep't of Corr., 641 F.3d 1276 (11th Cir. 2011); Moore v. Marr, 254 F.3d 1235 (10th Cir. 2001); Fugate v. Head, 261 F.3d 1206 (11th Cir. 2001); Jackson v. Herring, 42 F.3d 1350 (11th Cir. 1995); Nixon v. Newsome, 888 F.2d 112 (11th Cir. 1989)). Indeed, ”[h]armless-error analysis ... presupposes a trial, at which the defendant, represented by counsel, may present evidence and argument before an impartial judge and jury.” Rose, 478 U.S. at 578, 106 S.Ct. 3101 (emphases added). So the error that happens *1220when counsel is absent at trial when evidence is entered in error is different in quality and type than the kind that happens when counsel is present.
. While we may assess the effect of ineffective counsel, evaluating the effect of counsel’s absence when it rises to the level of Cronic error is another matter altogether. When counsel is present but allegedly ineffective, a transcript of what she did or did not do exists. So we can compare the choices she actually made against the broad spectrum of alternatives a reasonably competent attorney could have pursued. And we have a record of what counsel actually did during the entire trial — including not only how counsel responded to the erroneous admission of evidence but also how counsel conducted the rest of the trial after the error occurred. Put simply, we have the tools to allow us to evaluate the effects of the error on the actual defense, in light of a trial record created when the proceedings met the structural definition of a constitutional "trial.” As explained above, however, that is not the case when counsel in a single-defendant trial is absent. Comparison of Cronic and Strickland is also useful for another reason: it vividly demonstrates some of the other differences between the non-de minimis absent-counsel error that makes Cronic error structural error and the ineffective assistance of counsel that makes Strickland error trial error. In Strickland — significantly, issued on the same day as Cronic — the Supreme Court set up a dichotomy between cases involving the "[a]ctual or constructive denial of the assistance of counsel altogether,” see supra at 1215, which fall within Cronic's purview, and those where counsel was present throughout trial but arguably ineffective, which the teachings of Strickland govern. 466 U.S. at 692-93, 104 S.Ct. 2052. We presume prejudice in cases involving the "[ajctual or constructive denial of the assistance of counsel altogether” because prejudice is "so likely that case-by-case inquiry into prejudice is not worth the cost.” Id. at 692, 104 S.Ct. 2052. And violations in this group of cases "involve impairments of the Sixth Amendment right that are easy to identify and ... easy for the government to prevent.” Id. In contrast, cases where counsel was present but arguably ineffective concern alleged errors that the government is not responsible for, is often not able to identify while they are occurring, and is not able to prevent. Id. at 693, 104 S.Ct. 2052. And unlike error arising from the absence of counsel, alleged errors of ineffective assistance "cannot be classified according to likelihood of causing prejudice” because they come in so many varieties. Id. As the Court explained, "an act or omission that is unprofessional in one case may be sound or even brilliant in another.” Id. As a result, ineffective assistance cannot be defined precisely enough to put defense attorneys on notice of the conduct to avoid. Id. But the court and the prosecution can easily identify and avert the taking of directly inculpatory evidence in the absence of defense counsel in a single-defendant trial by simply electing not to proceed without defense counsel present.
. J.M. Barrie, Peter Pan, http://www. Iiteratureproject.com/peter-pan/peter-pan_l. htm (last visited Apr. 13, 2017) (on file with the Eleventh Circuit Clerk's Office).
. See Back to the Future (1985); Back to the Future Part II (1989); Back to the Future Part III (1990). As Christopher Lloyd’s character Dr. Emmett Brown explained in describing *1222the range of possible consequences/ going back in time could have virtually no effect on future events, or it could "cause a chain reaction that would unravel the very fabric of the space time continuum, and destroy the entire universe!” Back to the Future Part II, as quoted at http://www.imdb.com/title/tt0096874/ quotes (last visited Apr. 13, 2017) (on file with the Eleventh Circuit Clerk’s Office). For a more technical explanation of the space-time continuum, see https://einstein.stanford.edu/ content/relativity/q411 .html (last visited Apr. 13, 2017) (on file with the Eleventh Circuit Clerk’s Office).
. For this reason, the Majority opinion’s invocation of United States v. Noriega, 117 F.3d 1206 (11th Cir. 1997), see Maj. Op. at 1159-60, actually bolsters the point. Even in seven months of trial, the opinion does not indicate that counsel was absent for a single second of the taking of directly inculpatory evidence.
. Judge Wilson and Judge Martin take issue with drawing a precise, numerical line between trial error and structural error. That is a fair point. But Cronic expressly calls for a probability assessment. And the mere fact that the precise place to draw the line between the two types of error may not be immediately obvious does not mean that a category of absence that constitutes only trial error does not exist. We account for the lack of a readily discernible cutoff by drawing a line that necessarily includes all absences that could fairly be characterized as causing a breakdown of the adversarial process as doing so, even though it will also include some absences that do not so qualify on that side (e.g., 11 minutes in an 8-month trial). The cost of overinclusion in the structural-error category is, as a practical matter, extremely low, given the fact that holding trial without defense counsel in a single-defendant case almost never happens. See supra at 1222-24.
. Of course, the Sixth Amendment entitles each defendant to his own counsel. And where counsel labors under an actual conflict of interest at trial, that circumstance constitutes structural error. See Satterwhite, 486 U.S. at 256-57, 108 S.Ct. 1792 (citing Holloway, 435 U.S. at 490-91, 98 S.Ct. 1173). But many times, multiple defendants’ defenses are not inconsistent with one another. In addition, counsel sometimes agree to cover for each other with their clients’ permission. When these conditions exist, the breakdown in the trial process that might occur otherwise in a single-defendant trial simply does not occur.
. Sometimes a defendant chooses to be represented by more than one attorney.