This is Humberto Martin's third visit to this court. Four years ago, we affirmed his conviction for cocaine offenses. 964 F.2d 714 (7th Cir.1992). Two years ago, Martin filed in the district court a document that he styled a request for production of documents under the Jencks Act, 18 U.S.C. § 3500. The district court denied this and we affirmed in an unpublished order, remarking that the disposition of this irregular request was without prejudice to whatever relief might be available under 28 U.S.C. § 2255. Taking the hint, Martin filed such a petition, which the district court denied on the ground that the contentions have been forfeited by their omission from the direct appeal, are not cognizable under § 2255, or both.

With the aid of appointed counsel, Martin now presents three principal arguments (more properly, groups of arguments, but we needn't try to distinguish more finely). His double jeopardy argument has been authoritatively resolved against him. United States v. Ursery, --- U.S. ----, 116 S. Ct. 2135, 135 L. Ed. 2d 549 (1996). His argument under the Jencks Act is not properly presented. Both Martin and his appointed lawyer appear to believe that the Jencks Act is based on the Constitution, but it is not. United States v. Augenblick, 393 U.S. 348, 89 S. Ct. 528, 21 L. Ed. 2d 537 (1969); Jones v. DeRobertis, 766 F.2d 270 (7th Cir.1985). Martin also invokes Brady v. Maryland, 373 U.S. 83, 83 S. Ct. 1194, 10 L. Ed. 2d 215 (1963), but this record does not offer any reason to believe that the interview notes in question contain exculpatory evidence. Current counsel's critique of the way trial counsel phrased his requests for evidence is unavailing; it is not possible to show ineffective assistance without prejudice, Strickland v. Washington, 466 U.S. 668, 104 S. Ct. 2052, 80 L. Ed. 2d 674 (1984), and the lack of any support for the underlying claim makes that showing impossible.

What remain are a variety of objections to the calculation of the sentence. We held in Scott v. United States, 997 F.2d 340 (7th Cir.1993), that arguments based on the Sentencing Guidelines must be raised on direct appeal or not at all. (The exception for a fundamental miscarriage of justice is not activated by this case.) Martin tries to avoid Scott by arguing that his lawyer's failure to pursue sentencing issues on appeal was ineffective assistance of counsel. Yet in Durrive v. United States, 4 F.3d 548 (7th Cir.1993), we held that, to satisfy the "prejudice" component of Strickland, a person who complains about his sentence must demonstrate that counsel's deficient performance led to a "significant" increase in the sentence.

Although we recognized in Durrive that any extra time in prison is significant from the defendant's perspective, we adopted a rule of proportionality: the sort of increase produced by a few levels' difference in sentencing calculations cannot be raised indirectly on collateral attack by complaining about counsel's work. A few levels is exactly what Martin wants us to examine. Actually, his complaint is about a single criminal-history level (equivalent to two offense levels). Such an argument is barred by Scott when presented directly, and by Durrive when presented as an attack on counsel's performance. Martin asks us to disregard Durrive because the prosecutor did not cite it, but a court is not restricted to the cases the parties discuss. Elder v. Holloway, 510 U.S. 510, 114 S. Ct. 1019, 127 L. Ed. 2d 344 (1994).

AFFIRMED.

ON PETITION FOR REHEARING AND SUGGESTION OF REHEARING EN BANC

March 19, 1997.

Humberto Martin filed a petition for rehearing and suggestion of rehearing en banc on December 18, 1996. All members of the panel voted to deny the petition for rehearing. A judge in active service called for a vote on the suggestion of rehearing en banc, but a majority of the active judges voted to reject the suggestion. The petition for rehearing is denied, and the suggestion of rehearing en banc is rejected.

ILANA DIAMOND ROVNER, Circuit Judge, with whom RIPPLE and DIANE P. WOOD, Circuit Judges, join, dissenting from the denial of rehearing en banc.

Humberto Martin alleges that his appellate counsel was ineffective in failing to challenge a sentencing error that resulted in a sixteen-month increase in his sentence. In its opinion, the panel declined to consider whether error indeed did occur in application of the Sentencing Guidelines, as Martin argues. It found instead that Durrive v. United States, 4 F.3d 548 (7th Cir.1993), foreclosed collateral review of this claim, because the increase in Martin's sentence was not "significant" enough to satisfy the prejudice prong of Strickland v. Washington, 466 U.S. 668, 104 S. Ct. 2052, 80 L. Ed. 2d 674 (1984).

Today the full court regrettably declines to reexamine the holding in Durrive. I joined the opinion in Durrive, but upon reflection I have come to believe that we were mistaken in concluding that an increase in a prison term resulting from an unchallenged sentencing error does not constitute "prejudice" for purposes of the Strickland analysis unless the increase strikes us as "significant." Although Durrive's holding purports to follow from the Supreme Court's decision in Lockhart v. Fretwell, 506 U.S. 364, 113 S. Ct. 838, 122 L. Ed. 2d 180 (1993), and the Fifth Circuit's opinion in Spriggs v. Collins, 993 F.2d 85 (5th Cir.1993) (per curiam), the court in fact has both misconstrued Lockhart and failed to recognize the express limitations that the Fifth Circuit placed on the "significance" test we borrowed from Spriggs. The rule we adopted in Durrive is thus one wholly of our own making, and I do not believe it can withstand close scrutiny. When it is clear that an error has been made in applying the Sentencing Guidelines, and when it is clear that the error would have been corrected on direct review had the petitioner enjoyed effective advocacy, we are bound to recognize a violation of the defendant's Sixth Amendment right to counsel, notwithstanding our own view as to the "significance" of the additional prison time occasioned by the error. I do not see, in any event, how we can distinguish in any meaningful way increases in prison terms that are "significant" from those that are not. I must therefore dissent from the decision not to rehear this case en banc.

In 1990, Martin was convicted of conspiring to distribute cocaine, and his term of imprisonment was determined using the Sentencing Guidelines. In the presentence report, the probation officer concluded that Martin should be assigned a criminal history category of III, and the parties lodged no objection to that recommendation. Had the court employed that criminal history category, the resulting guidelines sentencing range would have been 97 to 121 months. At sentencing, however, the district court sua sponte assigned Martin a criminal history category of IV, which boosted the sentencing range to 121 to 137 months. The district court ordered Martin to serve the maximum sentence within that range (137 months), a term sixteen months longer than the high end of the range that a criminal history category of III would have produced.

Martin appealed directly to this court and we affirmed his conviction. United States v. Martin, 964 F.2d 714 (7th Cir.1992). As our opinion in that appeal reflects, no contention of error was raised as to the propriety of his sentence. And because our opinion in Scott v. United States, 997 F.2d 340 (7th Cir.1993), holds that a sentence resulting from an erroneous application of the Sentencing Guidelines is not a sentence "imposed in violation of the ... laws of the United States" (28 U.S.C. § 2255), collateral review of the error (if error there was) is foreclosed.

Having lost the sole opportunity to present his claim of sentencing error, Martin filed a section 2255 petition asserting that his appellate attorney was constitutionally ineffective for failing to raise the error on direct review. Any reasonable attorney would have argued this point on appeal, Martin posited, and the omission resulted in obvious and concrete prejudice: a sentence sixteen months longer than it would have been had the Sentencing Guidelines been applied correctly. Unfortunately for Martin, however, that claim ran headlong into our opinion in Durrive.

We held in Durrive that even when it is clear that an attorney's unreasonable mistake or omission has resulted in an erroneous sentence, the difference in outcome is insufficient to establish prejudice under the Strickland analysis; only if it is likely that the defendant's sentence would have been " 'significantly less harsh' " can there be a viable claim of ineffectiveness. 4 F.3d at 550 (quoting Spriggs v. Collins, 993 F.2d at 88) (emphasis in Spriggs). The springboard for that conclusion was Lockhart v. Fretwell, in which the Supreme Court observed:

[A]n analysis focussing solely on mere outcome determination, without attention to whether the result of the proceeding was fundamentally unfair or unreliable, is defective. To set aside a conviction or sentence solely because the outcome would have been different but for counsel's error may grant the defendant a windfall to which the law does not entitle him.

506 U.S. at 369-70, 113 S. Ct. at 842-43 (footnote omitted) (quoted in Durrive, 4 F.3d at 551). We construed Lockhart to mean that an error that produces a lengthier prison term does not make the sentence "unreliable or ... fundamentally unfair." 4 F.3d at 551. "Grave errors by judge and counsel"--"[f]or instance, an error that produces a large effect on the sentence"--could be deemed prejudicial for purposes of Strickland. Id. (emphasis supplied). But errors resulting in an extra "two or three steps" in the offense level do not fall into this category. Id. In Durrive, for example, the petitioner contended that had his attorney performed effectively, his offense level would have been two steps lower and his sentence at least one year shorter. See id. We conceded that "an extra year in custody is significant from the defendant's perspective." Id. "But the difference between 120 months and 108 or even 98 ... does not demonstrate that the actual sentence is 'unreliable ... or fundamentally unfair'--is not, in the language of Spriggs, a 'significant' difference." Id.

Although Durrive held out Lockhart as the foundation for its holding, I believe the court overlooked not only the context of the Supreme Court's decision but limitations in the very language of its opinion. The question before the Court in Lockhart was whether a habeas petitioner could prevail on an ineffectiveness claim based on his attorney's failure to invoke at sentencing a decision of the Eighth Circuit that was later overruled. An Arkansas jury had convicted the petitioner in Lockhart, Bobby Ray Fretwell, of felony murder and sentenced him to death. The jury found as an aggravating factor warranting the death penalty the fact that the murder (which had occurred during a robbery attempt) had been committed for pecuniary gain. Shortly before Fretwell was convicted and sentenced, however, the Eighth Circuit held in Collins v. Lockhart, 754 F.2d 258 (8th Cir.), cert. denied, 474 U.S. 1013, 106 S. Ct. 546, 88 L. Ed. 2d 475 (1985), that a death sentence could not constitutionally be premised upon an aggravating factor that is also an element of the felony of which the petitioner was convicted. Fretwell appealed his death sentence based on Collins, but the Arkansas Supreme Court declined to reach the merits of his argument because he had not made the objection during the sentencing phase of his trial. Ultimately, Fretwell brought his challenge to federal court, arguing that his trial counsel was ineffective in failing to invoke Collins. The district court vacated Fretwell's death sentence and ordered that he be given a new sentencing hearing. A divided panel of the Eighth Circuit agreed that relief on the ineffective assistance claim was warranted, despite the fact that it had since overruled Collins in light of intervening Supreme Court precedent. The majority reasoned that because Collins represented the law of the circuit at the time Fretwell was tried, the trial court would have been compelled to follow it had Fretwell's counsel made a timely objection and as a result the jury would not have sentenced him to death. Consequently, the majority believed itself obligated not simply to require a new sentencing hearing--at which Fretwell would be unable to invoke the overruled Collins--but to modify Fretwell's sentence outright to one of life imprisonment. To do otherwise "would perpetuate the prejudice caused by the original sixth amendment violation." Fretwell v. Lockhart, 946 F.2d 571, 578 (8th Cir.1991). The dissent thought it a mistake, in evaluating the prejudice resulting from an attorney's ineffectiveness, to focus exclusively on the probable difference in the outcome of the sentencing proceeding. Id. at 579. The Supreme Court agreed.

The Supreme Court emphasized that a single-minded focus on "mere outcome determination," without considering whether the result of the proceeding was unfair or unreliable, is a "defective" approach to evaluating attorney ineffectiveness, for it may result in an unwarranted "windfall" to the defendant. 506 U.S. at 369-70, 113 S. Ct. at 842-43. The Court cited Nix v. Whiteside, 475 U.S. 157, 106 S. Ct. 988, 89 L. Ed. 2d 123 (1986), to demonstrate the point. There the defendant contended that his attorney had been ineffective in refusing to elicit perjured testimony that might have exonerated him in the eyes of the jury. The Court rejected that argument, recalling Strickland's observation that "[a] defendant has no entitlement to the luck of a lawless decisonmaker." Lockhart, 506 U.S. at 370, 113 S.Ct. at 843 (internal quotation marks and citations omitted). The Court saw Fretwell's case as no different, given that his ineffectiveness claim was based on a case that was no longer good law. In retrospect, his attorney's failure to invoke that case had not undermined the fairness or the reliability of the sentencing proceeding; at most it had " 'deprived [Fretwell] of the chance to have the state court make an error in his favor.' " Id. at 371, 113 S.Ct. at 843 (quoting the amicus brief of the United States at 10).

As the Court's own choice of words suggests, it was concerned that the defendant not be granted "a windfall to which the law does not entitle him." Lockhart, 506 U.S. at 370, 113 S.Ct. at 843. Thus, what is clear from Lockhart is that the fairness and reliability of a proceeding, which is the touchstone for the Strickland ineffectiveness inquiry (see id. at 368-69, 113 S.Ct. at 842), cannot be judged in light of now-defunct legal precedents, even if the outcome of the proceeding would have been different had the defendant's attorney taken advantage of them. That holding flows directly from Strickland itself, which observed that "[a]n assessment of the likelihood of a result more favorable to the defendant must exclude the possibility of arbitrariness, whimsy, caprice, 'nullification,' and the like." 466 U.S. at 695, 104 S.Ct. at 2068 (quoted in Lockhart, 506 U.S. at 373, 113 S.Ct. at 845) (O'Connor, J., concurring). Lockhart's holding therefore is, as Justice O'Connor explained in her concurrence, "narrow." 506 U.S. at 374, 113 S.Ct. at 845.

Today the Court identifies another factor that ought not inform the prejudice inquiry. Specifically, today we hold that the court making the prejudice determination may not consider the effect of an objection it knows to be wholly meritless under current governing law, even if the objection might have been considered meritorious at the time of its omission. That narrow holding, of course, precisely disposes of this case as it appeared before the Eighth Circuit. The omitted objection of which respondent complained very well may have been sustained had it been raised at trial. But by the time the Eighth Circuit reviewed respondent's ineffective assistance claim, on-point Circuit authority bound that court to hold the objection meritless; the Arkansas Supreme Court had rejected the objection as well. Perry v. Lockhart, 871 F.2d 1384, 1392-1394(CA8), cert. denied, 493 U.S. 959, 110 S. Ct. 378, 107 L. Ed. 2d 363 (1989); O'Rourke v. State, 295 Ark. 57, 63-64, 746 S.W.2d 52, 55-56 (1988). Consequently, respondent's claim of prejudice was based not on the allegation that he was denied an advantage the law might permit him. It was predicated instead on the suggestion that he might have been denied "a right the law simply does not recognize," Nix, supra, 475 U.S. at 186-187, 106 S. Ct. at 1004 (BLACKMUN, J., concurring in the judgment), namely the right to "have the state court make an error in his favor," ante, at 371, 113 S. Ct. at 843 (opinion of the Court) (internal quotation marks omitted). It seems to me that the impact of advocating a decidedly incorrect point of law, like the influence of perjured testimony, is not a proper consideration when assessing "the likelihood of a result more favorable to the defendant." Strickland, supra, 466 U.S. at 695, 104 S. Ct. at 2068. I therefore join the Court in holding that, in these somewhat unusual circumstances, the Court of Appeals should have concluded that respondent suffered no legally cognizable prejudice.

506 U.S. at 374-75, 113 S. Ct. at 845-46. So understood, the decision in Lockhart "will, in the vast majority of cases, have no effect on the prejudice inquiry under Strickland[.]" Id. at 373, 113 S. Ct. at 845.

What quite obviously sets this case apart from Lockhart is that Martin is not relying on a "decidedly incorrect statement of the law" or anything else that falls into the category of "arbitrariness, whimsy, caprice, 'nullification,' and the like" to establish prejudice. He is not seeking a "windfall." What Martin contends, as the petitioner in Durrive contended, is that for no legitimate strategic reason, his attorney failed to make an argument that the law recognized then and still recognizes now, and which, had it been made, would have resulted in a lesser sentence. Absolutely nothing in Lockhart addresses this situation. Nowhere does Lockhart suggest that a difference in outcome may be overlooked when it results from the failure to make arguments that were and remain valid under governing legal principles. Nowhere does Lockhart suggest that a sentencing may still be described as fair and reliable when it is clear that the sentencing court has sentenced the petitioner to an excessive term, and only his attorney's ineffectiveness has prevented correction of the error either in the district court or on direct appeal. Nothing in Lockhart remotely suggests that only when an attorney's ineffectiveness results in a "significantly" harsher sentence may the prejudice be recognized on collateral review.

Durrive derived its "significance" test from the Fifth Circuit's opinion in Spriggs v. Collins, but as with its invocation of Lockhart, Durrive overlooked the particular circumstances that led that court to rule as it did. Waymon Spriggs had pleaded guilty to a charge of first-degree murder and a Texas state court had ordered him to serve a prison term of thirty-five years. Spriggs sought relief from the sentence pursuant to 28 U.S.C. § 2254, contending that his counsel at sentencing was ineffective for not objecting to the presentence investigation report's representation that he had a "long history of assaultive and aggressive behavior." The Fifth Circuit observed at the outset of its analysis that under a "rigid application" of Strickland, the prejudice prong--which requires a reasonable probability that but for counsel's error the outcome would have been different--might be relatively easy to meet in a non-capital sentencing case:

This is because a non-capital sentencer does not simply have the choice of life or death. Rather, non-capital sentencing hearings, particularly in jurisdictions without sentencing guidelines, typically involve wide sentencing discretion. In the case of non-capital murder in Texas, such discretion is extensive. See TEX.PENAL CODE § 12.32 (sentencing range for first degree felony, including murder, is from five to 99 years, with option of $10,000 fine). Arguably, when the discretionary sentencing range is great, practically any error committed by counsel could have resulted in a harsher sentence, even if only by a year or two.

993 F.2d at 88. Thus, it was "[i]n order to avoid turning Strickland into an automatic rule of reversal in the noncapital context" that the court adopted a rule requiring a reasonable probability that the sentence would have been "significantly less harsh" in order to satisfy the prejudice prong of Strickland. Id. (emphasis in original). Yet, a careful reading of the court's opinion reveals that it was adopting a rule only for the case in which the sentencing court enjoyed wide discretion in selecting a punishment and in which there is a commensurate difficulty in identifying with particularity the impact that the attorney's error may have had on the sentence. Thus, immediately following its articulation of the "significantly less harsh" standard, the court observed in a footnote:

We note that one foreseeable exception to this requirement would be when a deficiency by counsel resulted in a specific, demonstrable enhancement in sentencing--such as an automatic increase for a "career" offender or an enhancement for use of a handgun during a felony--which would have not occurred but for counsel's error.

Id. at 89 n. 4 (emphasis supplied). By its own terms, then, Spriggs leaves the door open to relief in cases where the ineffectiveness of sentencing counsel results in a concrete, readily ascertainable increase in the defendant's prison term, even if the increase is not "significant."

In purporting to follow Spriggs' lead, we overlooked the express qualifications in the Fifth Circuit's own rationale. Where a petitioner like Martin alleges that his attorney's ineffectiveness either contributed to a misapplication of the Sentencing Guidelines (by failing to object to an evident error in the presentence report, for example) or left that error uncorrected (by failing to raise the error on direct appeal), there typically will be no need to speculate as to the impact of the error on the petitioner's sentence. The Guidelines themselves, which strive for uniformity of punishment through the confinement of judicial discretion in sentencing, make it possible to quantify the impact that the attorney's error had upon the petitioner's sentence. We are presented with what the Fifth Circuit described as a "specific, demonstrable enhancement in sentencing," one which negates the need for the "significantly less harsh" standard that court articulated. 993 F.2d at 89 n. 4. Yet, without so much as a nod to that distinction, Durrive simply lifted the "significantly less harsh" language from Spriggs and applied it wholesale to any and all sentencing errors attributable to an attorney's ineffectiveness, including errors that have a readily discernable impact on the sentence.

We have thus transformed Spriggs into an ostrich of a rule that deliberately blinds us even to an error that produces a quantifiable increase in a petitioner's sentence. Nothing in Spriggs supports that result; on the contrary, if we are to take the language of Spriggs seriously, the Fifth Circuit itself would not countenance that outcome. Without precedential support, then, we have fashioned a rule that without exception requires a petitioner to convince us that the excess in his prison term is "significant" before we will entertain the prospect of relief for his attorney's ineffectiveness.

Exactly what does constitute a "significant" increase in someone's incarceration? We know from Durrive that an increase of twelve months is not "significant," and from this case that sixteen months is not. Durrive suggests that a twenty-two month increase is also not "significant." See 4 F.3d at 551. Indeed, Durrive suggests that any increase resulting from an error of "two or three steps" in calculation of the offense level is not significant enough to warrant relief on collateral review. Id. A perusal of the sentencing table reveals that two or three steps in the offense level can mean eight or more additional years in prison. For someone in a criminal history category of V, for example, a final offense level of 34 calls for a prison term of 235 to 293 months, whereas an offense level three steps higher requires a term of 324 to 405 months--eighty-nine more months at the low end of the range, 112 months more at the high end. See United States Sentencing Commission, Guidelines Manual, Ch. 5, Pt. A, Sentencing Table (Nov.1995). If an increase of eight or nine years is not "significant," what is? Does the "significance" depend on the overall length of the petitioner's term? In other words, would an increase of a year be "insignificant" for someone serving a ten-year term, but "significant" for someone serving a two-year term?

Durrive leaves such questions unanswered, freeing individual judges and panels to reach their own conclusions. And it already appears clear that the results will be inconsistent. In United States v. Kissick, 69 F.3d 1048 (10th Cir.1995), cert. denied, --- U.S. ----, 117 S. Ct. 1008, 136 L. Ed. 2d 886 (1997), for example, the court embraced the approach we took in Durrive but found that an attorney's ineffectiveness had, in fact, contributed to a "significant" increase in the petitioner's sentence. The petitioner had been deemed a career offender based in part upon an apparently erroneous classification of one of his prior convictions. His attorney had not objected to the classification either at sentencing or on direct appeal, and the court (assuming that the classification was indeed erroneous) found that failure to be unreasonable. Id. at 1056. When the court turned to the prejudice prong of the Strickland analysis, it noted that once the petitioner had been found a career offender, his criminal history category jumped from IV to VI, and the possible imprisonment rose from a range of 262 to 327 months to a range of 324 to 405 months. The district court had sentenced Kissick to 328 months, so the increase attributable to the unchallenged sentencing error was at least one but not more than sixty-six months. In the Tenth Circuit's view, that increase was "significant," the prejudice "clear." Id. Yet, like our own opinion in Durrive, the Tenth Circuit's opinion in Kissick gives no guidance as to why an increase of one to sixty-six months is "significant" but an increase of twelve or twenty-two months is not. See also, e.g., United States v. Ford, 918 F.2d 1343, 1350 (8th Cir.1990) (trial counsel's failure to object to offense level that may have been two steps higher than it ought to have been, resulting in sentence at least thirty-eight months longer, sufficed to establish prejudice); Cabello v. United States, 884 F. Supp. 298, 302-03 (N.D.Ind.1995) (prejudice shown where sentence would have been at least eight years, eleven months shorter but for appellate counsel's failure to challenge error in assessment of petitioner's criminal history); Smith v. United States, 871 F. Supp. 251, 256 (E.D.Va.1994) (prejudice shown even where petitioner's sentence of 292 months was well within Guidelines range of 262 to 327 months that would have applied had attorney caught error in determination of criminal history category); United States v. Wolf, 1992 WL 129640, * 6 (E.D.Pa. June 8, 1992) (prejudice established even where petitioner's sentence of 132 months fell within range of 110 to 137 months that would have been used had counsel objected to error in calculation of relevant drug quantity).

I cannot help but believe that in the adoption of this "significance" test, we have transformed ourselves into a kind of feudal duchy, with the power to grant or deny a reprieve to wrongfully sentenced prisoners according to our own whim. This is repugnant not only to our office as appellate judges but to the Guidelines themselves, which after all were adopted to achieve greater consistency in sentencing among similarly situated defendants.

When this court concluded in Scott v. United States, supra, that errors in the application of the Sentencing Guidelines cannot be raised on collateral review absent extraordinary circumstances, we observed:

Congress provided for enforcement of the Guidelines by an adversarial clash, and in this sense they create "rights." Still, the Sentencing Guidelines are not exactly a Bill of Rights for criminals! One full and fair opportunity to make arguments under the Guidelines--at sentencing and on direct appeal--is enough.

997 F.2d at 342. This view presupposes that the defendant has the effective assistance of an attorney, for without such assistance there is no true "adversarial clash," no "full and fair opportunity" to make the types of arguments--often quite technical and detailed, given the nature of the Guidelines--that must be made to ensure fair and accurate sentencing. But the premise of Martin's claim is that he has been denied this assistance, that although he was represented by counsel, his attorney failed to pursue correction of an obvious error in the calculation of his sentencing range. See United States v. Ford, supra, 918 F.2d at 1350 (trial counsel's failure to object to incorrect offense level "was so serious that it deprived Ford of a fair sentence"); Smith v. United States, supra, 871 F.Supp. at 255 ("While courts are properly deferential to attorneys' discretion, failure to raise an objection to a clear and indisputable error in the PSR is not within the broad range of performance that can be deemed reasonable."). If Martin is correct in his assertion that the district court erred in its selection of a higher criminal history category, then his appellate counsel omitted the type of appellate issue that Judge Easterbrook has described as a "dead-bang winner." Page v. United States, 884 F.2d 300, 302 (7th Cir.1989). Yet, although the court is willing to assume that the error may have been overlooked, to Martin's obvious detriment, it declines to reach the merits of the issue because the additional sixteen months that Martin must serve in prison, although concededly significant to him (see ante at 1178) is not "significant" in the eyes of the court.

What the court has effectively done, it seems to me, is to lower dramatically the quality of advocacy vis a vis sentencing that will satisfy the Sixth Amendment. For although the court is willing to assume that a sentencing error was made and left unaddressed as the result of the ineffectiveness of Martin's counsel, and it is willing to assume that the error increased Martin's sentence, it refuses to entertain the prospect of relief. If there is no redress for the denial of effective assistance of counsel, then practically speaking there is no real right to such assistance. In Durrive's language, only a "grave error" producing a "large effect" on the sentence might support relief on collateral review. 4 F.3d at 551. That language, it would appear, sets the bar for the quality of advocacy that a defendant can expect when it comes to sentencing issues: the defendant is entitled only to an attorney who does not overlook "grave errors" having a "large effect" on the sentence. I am less than confident that this can be reconciled with the standard that the Supreme Court articulated in Strickland. And with the benefit of hindsight, I am certain that in joining Durrive I was wrong.