Randall J. Cofske, Petitioner, Appellant, v. United States of America, Respondent, Appellee, 290 F.3d 437 (1st Cir. 2002)Annotate this Case
Heard April 2, 2002
Decided May 13, 2002
Robert J. Bray for petitioner.
Randall J. Cofske on Motion for Emergency Hearing pro se.
Donald L. Cabell, Assistant United States Attorney, with whom Michael J. Sullivan, United States Attorney, was on brief for respondent.
Before BOUDIN, Chief Judge, BOWNES, Senior Circuit Judge, and LIPEZ, Circuit Judge.
BOUDIN, Chief Judge.
Randall Cofske appeals from the district court's denial of his motion under 28 U.S.C. § 2255 to vacate, set aside, or correct his sentence. In substance, Cofske claims that his trial counsel and appellate counsel were ineffective in failing to object to the calculation of his sentence under the Sentencing Guidelines. Specifically, he argues that when he committed the federal offense for which he was being sentenced — possession of stolen firearms in violation of 18 U.S.C. § 922(j) (1994) — he did not have the "prior conviction" that was used to increase his base offense level ("BOL") under U.S.S.G. § 2K2.1 (a) (4) (1995).
Cofske committed the federal offense in 1992, but we start with the earlier state offense that lies at the heart of his present appeal. On October 20, 1989, Cofske was charged in the Ware District Court in Massachusetts with breaking and entering in the daytime with the intent to commit a felony. Under the two-tier trial system then used by Massachusetts, Cofske waived his right to a jury trial and instead elected a bench trial in the Ware District Court (the first tier).1 On August 17, 1990, the judge found Cofske guilty of the charge and sentenced him to 60 days in prison. Cofske then appealed, seeking a trial de novo (the second tier) before a jury of six; under Massachusetts law this appeal vacated the initial conviction. Wilson v. Honeywell, Inc., 409 Mass. 803, 569 N.E.2d 1011, 1015 (1991). He was convicted again by a jury in the Northampton District Court on January 23, 1993.
After Cofske's conviction and appeal in Ware, but before his conviction in Northampton, Cofske committed the federal firearms offense. On or about October 27, 1992, Cofske and two other men broke into a Salem, Virginia, residence and stole about ten handguns. Using Cofske's car, they then transported the guns to Massachusetts where they were later sold. The grand jury indicted Cofske on July 17, 1996, for possession of stolen firearms in violation of 18 U.S.C. § 922(j) and for transportation of stolen firearms in violation of 18 U.S.C. § 922(i).
On March 3, 1997, pursuant to a written plea agreement, Cofske pleaded guilty to the charge of possession of stolen firearms. As part of that agreement the prosecution dismissed the transportation-of-stolen-weapons charge and agreed to recommend a single three-level specific offense adjustment (based on the number of firearms involved) offset by a three-level reduction for acceptance of responsibility; Cofske in turn agreed (among other things) to take the position at sentencing that his base offense level should be set at 20 pursuant to U.S.S.G. § 2K2.1(a) (4) (a), which mandated a BOL of 20 for firearms possession if "the defendant had one prior felony conviction of either a crime of violence or a controlled substance offense."2
The original presentence report ("PSR") listed Cofske's BOL as 12, concluding that he had no such prior conviction. After the government objected, the probation officer amended the PSR to include the Massachusetts breaking and entering conviction already described and increased his BOL to 20. Cofske's attorney urged in turn that the breaking and entering charge did not become a conviction until January 23, 1993 (at the second tier), which was after the federal offense, and therefore could not be counted as a prior conviction under U.S.S.G. § 2K2.1. The probation officer responded that the 1990 first tier conviction was the relevant prior conviction, and the PSR left the BOL at 20.
Cofske was sentenced on July 15, 1997. At sentencing, Cofske's attorney abandoned the argument that Cofske did not have a prior conviction and stated that the revised PSR "took up acknowledgment of the government's objection which was quite correct and part of my error at the time. It starts out with a base level of 20 pursuant to 2K2.1(a) (4)." The district court then adopted the calculations of the revised PSR, which included the breaking and entering conviction in both Cofske's criminal history category ("CHC") calculation under section 4A1.1, and his BOL under section 2K2.1.
The court determined that Cofske's BOL was 20, his adjusted offense level was 24,3 and his CHC was V; the applicable guideline sentence range was 92 to 115 months. After denying Cofske's motion for a downward departure, the district court sentenced Cofske to 92 months in prison. Cofske appealed — not raising the issue he raises here — and this court affirmed both his conviction and sentence. United States v. Cofske, 157 F.3d 1 (1st Cir. 1998), cert. denied, 526 U.S. 1059, 119 S. Ct. 1374, 143 L. Ed. 2d 533 (1999).
Cofske then filed a motion under 28 U.S.C. § 2255 asserting that, at the time of his federal firearms offense, he had not been convicted of the breaking and entering offense because his 1990 first-tier conviction in the Ware District Court had been nullified by his appeal seeking a second-tier de novo jury trial. He further argued that this conviction was impermissibly used to enhance his BOL and CHC, and that his trial counsel and appellate counsel were ineffective for failing to object to its inclusion in his sentence computation.
On May 18, 2000, the district court denied Cofske's motion. The court found that the 1990 first-tier conviction was the relevant conviction for purposes of section 2K2.1(a) (4) (A) and that there was no error in the calculation of Cofske's sentence; it determined that Cofske was therefore not prejudiced by his counsel's actions, as required by Strickland v. Washington, 466 U.S. 668, 104 S. Ct. 2052, 80 L. Ed. 2d 674 (1984), and found it unnecessary to evaluate the reasonableness of Cofske's attorney's decision not to object to the BOL calculation. The district court later denied Cofske's requests for reconsideration of the decision.
On February 8, 2001, Cofske applied for a certificate of appealability, 28 U.S.C. § 2253(c) (1), solely to urge that his trial and appellate counsel were ineffective in allowing his first-tier conviction to increase his BOL. On April 20, 2001, the district court denied his application, reaffirming its May 18, 2000, finding and also holding (in the alternative) that the breaking and entering conviction counted as a "prior conviction" even if the correct date of conviction were deemed January 1993. Cofske v. United States, 143 F. Supp. 2d 85 (D. Mass. 2001).
On November 26, 2001, this court granted Cofske's certificate of appealability and directed the parties to address the following two questions:
1. Under the Massachusetts two-tier trial court system in effect at the time of Cofske's conviction, what is the force and effect of a district court conviction once the defendant has appealed to a trial de novo?
2. Does the reference to a "prior felony conviction" in section 2K2.1(a) (4) (A) include post-offense convictions?
We also granted Cofske's request for expedited briefing, because on his calculation, his proper sentence had already expired.
Although the language of 28 U.S.C. § 2255 is quite general, the Supreme Court has restricted collateral attack for claims that do not allege constitutional or jurisdictional errors; such claims are said to be cognizable only where the alleged error presents "a fundamental defect which inherently results in a complete miscarriage of justice" or "an omission inconsistent with the rudimentary demands of fair procedure." Hill v. United States, 368 U.S. 424, 428, 82 S. Ct. 468, 7 L. Ed. 2d 417 (1962). Thus, a guideline violation alone is not automatically a basis for relief under 28 U.S.C. § 2255. Knight v. United States, 37 F.3d 769, 772-73 (1st Cir. 1994).
However, if the claim is repackaged as one of ineffective assistance of counsel, as Cofske's is here, it becomes a constitutional claim. Not every error amounts to ineffectiveness. See Lema v. United States, 987 F.2d 48, 51 (1st Cir. 1993). An ineffective assistance of counsel claim will succeed only if the defendant — who bears the burden on both points, Scarpa v. DuBois, 38 F.3d 1, 8-9 (1st Cir. 1994) — shows (1) that counsel's performance fell below an objective standard of reasonableness, and (2) that but for the error or errors, the outcome would likely have been different, Strickland, 466 U.S. at 687, 104 S. Ct. 2052.
Since the absence of any error in sentencing would eliminate any prejudice, and therefore Cofske's ineffectiveness claim, it is useful to begin by considering the correctness of the BOL calculation. At the time of Cofske's sentencing in July 1997, U.S.S.G. § 2K2.1(a) (4) (A) stated that a defendant's BOL for the firearms conviction should be set at 20 if the defendant "had one prior felony conviction of either a crime of violence or a controlled substance offense[.]" However, neither the guideline language nor the commentary answered the question, "prior to what?"
Certainly the guideline would be satisfied if the "prior conviction" occurred before the federal crime itself was committed. But on this reading, we would have to count the Ware District Court conviction as a prior conviction, even though it had been vacated as a result of Cofske's filing an appeal for a second tier trial, months before Cofske committed the federal offense.
Whether such a vacated conviction still "counts" for guideline purposes is a matter of federal rather than state law. See United States v. Mateo, 271 F.3d 11, 15 (1st Cir. 2001). Looking only at guideline language, one could argue the matter both ways. Before he committed the federal crime, Cofske "had one prior felony conviction" of the required class; but before he committed the federal crime, that conviction had been nullified under state law by his appeal from the first tier to the second. The commentary was understandably silent on how to treat this oddity of Massachusetts procedure.
There are policy arguments both ways. If the issue were pursued in depth, we might want to know more about the reasons for the two-tier regime and, equally important, whether it led lawyers to use the first tier merely as a discovery device or an effort to secure probation or a light sentence. Cf. United States v. Roberts, 39 F.3d 10, 12-13 (1st Cir. 1994). And, of course, whether the first tier conviction followed a full defense might vary from case to case. If in general a full defense were offered, federal law might well disregard the state's penchant to offer the defendant a second bite at the apple.
In all events, the government has chosen here not to argue that the Ware District Court conviction should count. The concession has arguments in its favor and the opposing arguments are debatable and have not been briefed. For purposes of this case, we accept the concession, which turns out not to affect the outcome here. Given the repeal of the two-tier system, it is not clear that the issue will even arise again so it certainly does not cry out for a definitive resolution.
This brings us to the district court's alternative ground for concluding that the prior conviction should count, namely, that even if the Northampton District Court conviction is controlling, it occurred "prior" to the federal sentencing, even though after the federal crime itself. Neither the version of the section 2K2.1 guideline in force in mid 1997 nor its commentary clearly answers the question whether the conviction must be prior to the federal crime or only prior to the sentencing for it.
In fact, the clues to this puzzle — linguistic, policy and analogical — point in both directions and, prior to a recent amendment to the guideline described below, the circuits were divided. At the time of the Cofske's federal sentencing, the circuits were split two to one in the government's favor (i.e., that any conviction prior to federal sentencing counted);4 at the time of the recent amendment, the government still held the edge with a four-to-three split.5 Obviously, the issue was close and reasonable courts could disagree.
Part of the problem is that the guidelines elsewhere contain two different approaches to "prior"-ness. One, used in determining prior criminal history under chapter 4A, looks to whether a prior "sentence" occurred before the federal sentence. U.S.S.G. § 4A1.2, application note 1. The other, used to establish whether the defendant is a career offender under chapter 4B, counts prior "convictions" only if they occurred before commission of the federal crime itself. U.S.S.G. § 4B1.2(3). In its own context, each approach makes some sense. United States v. Leviner, 31 F. Supp. 2d 23, 28-30 (D. Mass. 1998).
The issue has now been resolved for the future in Cofske's favor by a 2001 amendment to section 2K2.1 (reprinted in the addendum) providing that a defendant's BOL should be increased to 20 where "the defendant committed any part of instant offense subsequent to sustaining one felony conviction of either a crime of violence or a controlled substance offense." U.S.S.G. § 2K2.1(a) (4) (A) (2001) (emphasis added). The amendment's "reason for adoption" adopts "the minority view" and the chapter 4B analogy, Amendment 630, Nov. 1, 2001, but it was not made retroactive to Cofske's case. U.S.S.G § 1B1.10 (2001). One could as easily call it a revision as a clarification.
Without regard to the amendment, we think that prior convictions should count, for purposes of section 2K2.1 as it stood when Cofske was sentenced, only if they occurred before the federal crime. This was the "minority" view in the circuits prior to the recent guideline amendment, but it has in its favor a modest piece of linguistic evidence, Pedragh, 225 F.3d at 245, a decent analogy to the prior-felon "status" approach used in section 4B, Leviner, 31 F. Supp. 2d at 28-30, and a better claim to adoption by cross-reference in a tangled set of cross references in 2K2.1's then-existing commentary.6 Both sides of the dispute can be traced through the cited cases, see notes 4-5, above.
Still, our current view of the merits hardly establishes that Cofske's counsel was ineffective in 1997 in failing to press the argument for a BOL of 12. The guideline's original wording is obscure and a majority of circuits both before and after read the pre-amendment guideline in favor of the prosecutor. If Cofske's counsel had simply overlooked the argument, it is uncertain that this would have counted as incompetence, even though courts tend to be somewhat less forgiving where counsel altogether overlooks a possible objection or opportunity. 3 LaFave, Israel & King, Criminal Procedure § 11,10(c), at 714-15 (2nd ed.1999).
However, Cofske's trial counsel did not overlook the issue; the PSR resolved it in Cofske's favor and his counsel objected when the probation officer reversed ground and raised the recommended BOL to 20. At sentencing, Cofske's counsel then abandoned the objection. If counsel gave away without any cause a known argument which had at least some basis in guideline language and policy, this would at least take some explaining, especially where (as here) the jump in the BOL from 12 to 20 greatly increased the guideline sentence range.
Yet there is an obvious reason why counsel must have thought it in Cofske's interest to abandon the claim of error. In the plea agreement, the government had agreed (1) to drop the second count of the indictment, (2) to support a three-level reduction for acceptance of responsibility, (3) to seek an adjusted offense level of 20 (and not a higher level), and (4) not to seek any upward departure.7 Cofske in turn made concessions of his own; and these included not only a guilty plea on the first count but also a commitment to take the position at sentencing that the BOL "should be 20 ... because the defendant had one prior felony conviction for a crime of violence."
It is virtually certain that prior to the sentencing, defense counsel noticed, or was reminded of, this commitment. Nothing else explains counsel's about face at the hearing after previously contesting the issue. Any breach of the agreement by Cofske permitted the government to withdraw from the plea agreement and to use against Cofske any statements he had made. Conceivably, the government could have asked to reinstate the second count; and certainly, it could have urged any sentence it chose including one based on no acceptance of responsibility plus upward departures on at least two different grounds.
Thus, the decision to stick with the BOL of 20 was almost certainly a judgment call to retain the advantages of a plea and to forgo an argument for a lower BOL already rejected by two out of three circuits. It is impossible to describe such a choice as incompetent representation. Nagi v. United States, 90 F.3d 130, 134-35 (6th Cir. 1996). Indeed, given the threatened consequences to Cofske of backing out of the plea agreement, it is impossible to see how Cofske could now satisfy the prejudice prong of Strickland by showing that he would have been better off if his counsel had challenged the use of the state conviction.
If anything turned on counsel's precise thought process, we would remand for an evidentiary hearing, but in this case none is necessary. The Strickland test, as already noted, is an objective one; as long as counsel performed as a competent lawyer would, his or her detailed subjective reasoning is beside the point. Here, counsel was aware of the BOL objection and, given the downside risks already described, trial counsel had ample reason to comply at sentencing with commitments undertaken in the plea agreement, whose soundness (taken as a whole) is not even directly disputed by Cofske.
This brings us to Cofske's alternative ground for relief, namely that his appellate counsel was incompetent in not pursuing a claim to a BOL of 12 on his direct appeal. After all, at this point Cofske seemingly had the benefit of the government's concessions; why then not try to get something more — namely, a reduced BOL — even if the chances of success appeared slim? The plea agreement precluded this tactic at the sentencing hearing; but it did not bar either side from appealing. Indeed, the agreement provided (unnecessarily) that errors in sentencing could be appealed under 18 U.S.C. § 3742.
Of course, Cofske's concession in the district court that a BOL of 20 was correct might have been taken as a conclusive waiver, not just a forfeiture subject to relief for plain error, but then again it might not be treated as a waiver; the law in this area is far from uniform, compare United States v. Smith, 918 F.2d 664, 668-69 & n. 1 (6th Cir. 1990) (per curiam), with United States v. Nguyen, 46 F.3d 781, 783 (8th Cir. 1995), and various interpretations of the plea agreement are possible. Thus, it may technically have been open to appellate counsel to challenge the adverse BOL finding on direct appeal.
It might be enough to say that, viewed as of the time of appeal (which is what matters in judging competence of appellate counsel), the claim for a lower BOL was not especially promising: the case law was divided, the guideline and commentary were both obscure, and the 2001 amendment did not then exist. Other claims were available for an appeal, which was in fact taken, and courts have regularly said that appellate counsel is often well advised to choose the most promising arguments and is not obliged to crowd a brief with less promising ones which may detract. Jones v. Barnes, 463 U.S. 745, 751-54, 103 S. Ct. 3308, 77 L. Ed. 2d 987 (1983); see also Burkoff & Hudson, Ineffective Assistance of Counsel § 9.04[b] (Supp.2001) (collecting cases).
Here, a narrower and even stronger reason exists for rejecting an incompetency attack. On appeal, the claim to a BOL of 12 could succeed only if plain error were established; even without the plea agreement, a failure to press the BOL claim at sentencing was at least a forfeiture. However one might come out on the merits of the BOL issue, the BOL of 20 was not plain error for the reasons already indicated. For all we know, a panel of this court sitting in 1997 might easily have followed the then-majority view and said that 20 was the correct figure.
But there is more. Even if the error were plain, any competent appellate counsel would have had to consider not only the upside but also the downside of attacking the district court's BOL finding. At the very least, such an attack by Cofske would have exposed him to arguments by the government that — if resentencing were ordered — the resentencing should be open-ended and the government should be freed of its own commitments to support a downward adjustment for acceptance of responsibility and to forego various upward departures.
Once again, the law in this area is not crystal clear; just what the government might demand as of right, and (separately) what the appeals and sentencing courts might allow as a matter of discretion, could both be debated and could turn in some measure on the surrounding circumstances. However, Cofske is wrong in thinking that he necessarily had a right to cherry-pick, correcting the BOL in his favor while holding everything else in the sentencing calculus constant.8 An appellate counsel, even with a winning argument, might reasonably have hesitated to press it if the downside risk seemed too great.
For present purposes, it is enough that the BOL of 20 was not plain error and that this would almost certainly have defeated a direct appeal at the time in view of the forfeiture (also reasonable) at the sentencing stage. Given the minimal prospects of success, appellate counsel's failure to make such an argument on direct appeal does not mark counsel as incompetent. This is a legal judgment that we can easily make on this appeal and without the need for a remand.
Courts are driven as much by facts as by doctrine, and in criminal proceedings no panel of this court readily closes the books on a manifest injustice. Nothing of the kind is present here. The 12 versus 20 BOL issue was known to defense counsel at the time of sentencing; a rational choice was made to forego the argument for 12 in order to obtain other sentencing advantages. Even with the benefit of hindsight, there is no clear likelihood that a different choice by counsel would have given Cofske a lower net sentence.
PERTINENT PROVISIONS OF SECTION 2K2.1
At the time of sentencing
§ 2K2.1 Unlawful Receipt, Possession, or Transportation of Firearms or Ammunition: Prohibited Transactions Involving Firearms or Ammunition
(a) Base Offense Level (Apply the Greatest):
. . .
(4) 20, if the defendant —
(A) had one prior felony conviction of either a crime of violence or a controlled substance offense; or
. . .
(7) 12, except as provided below, or
. . .
(b) Specific Offense Characteristics
(1) If the offense involved three or more firearms, increase as follows:
Number of Firearms Increase in Level
(C) 8-12 add 3
. . .
After the 2001 amendment
. . .
(4) 20, if —
(A) the defendant committed any part of the instant offense subsequent to sustaining one felony conviction of either a crime of violence or a controlled substance offense; or
. . .
As explained in United States v. Morillo, 178 F.3d 18, 21-22 n. 5 (1st Cir. 1999):
Under this `two-tier' system, a defendant was entitled to a six person jury trial at the `first tier.' The defendant could waive this right and elect a bench trial instead. Defendants who elected a bench trial could appeal the outcome to the `second tier.' At this level, a defendant would receive a de novo jury trial unless he again elected a bench trial before a second judge.
In 1994, Massachusetts abolished the two-tier trial system in the Commonwealth's district courts.
References, unless otherwise indicated, are to the 1995 edition of the guidelines, as updated by a 1996 pamphlet. This is the set of guidelines in effect at the time of Cofske's sentencing in mid 1997. The pertinent provision of section 2K2.1 as of mid 1997 is set forth in an addendum to this opinion
The court imposed a three-level upward adjustment under § 2K2.1(b) (1) (C) because the offense involved between 8 and 12 firearms, and a four-level upward adjustment under § 2K2.1(b) (5) because Cofske possessed the firearms in connection with a felony offense or transferred them with reason to believe they would be used in another felony offense. The district court then applied a three-level reduction for acceptance of responsibility, thereby setting Cofske's adjusted offense level at 24. None of these adjustments is now challenged
Compare United States v. Gooden, 116 F.3d 721, 724-25 (5th Cir. 1997) and United States v. McCary, 14 F.3d 1502, 1506 (10th Cir. 1994) (convictions occurring anytime before sentencing count as prior felony convictions) with United States v. Barton, 100 F.3d 43, 45-46 (6th Cir. 1996) (post-offense convictions do not count as prior felony convictions).
United States v. Laihben, 167 F.3d 1364, 1366 (11th Cir. 1999), and United States v. Pugh, 158 F.3d 1308, 1311 (D.C. Cir. 1998) followed the approach of the Fifth and Tenth Circuits, while United States v. Oetken, 241 F.3d 1057, 1059 (8th Cir. 2001) and United States v. Pedragh, 225 F.3d 240, 245 (2d Cir. 2000), followed the approach of the Sixth Circuit.
Section 2K2.1, app. note 5, said that "prior felony conviction(s)" is defined in "§ 4B1.2... Application Note 1 of the Commentary" which in turn shed no light on the question before us but was at least a reference to chapter 4B rather than 4A; chapter 4B's approach, as noted in text, supports Cofske. Confusingly, application note 5 and application note 1 also contain cross-reference to chapter 4A, although arguably less in point
At oral argument, the government summarized the potential changes that it could have sought: an upward departure on the ground that past criminal history was underrepresented, U.S.S.G. § 4A1.3; a two-level obstruction of justice adjustment for false statements to the court, U.S.S.G. § 3C1.1; and a four-level adjustment under U.S.S.G. § 2K2.1(b) (5) for possessing or transferring firearms in connection with another felony offense. Although the court on its own imposed the section 2K2.1(b) (5) adjustment, the government did not seek it
On any of several theories, the government could have sought to be relieved of its obligations under the plea agreement,see United States v. Bunner, 134 F.3d 1000, 1004-5 (10 Cir. 1998); United States v. Barron, 127 F.3d 890, 895-96 (9th Cir. 1997); United States v. Sandoval-Lopez, 122 F.3d 797, 800 (9th Cir. 1997). Further, the district court would not automatically be limited on remand by what the government chose to urge. United States v. Rodriguez, 112 F.3d 26, 29-30 (1st Cir. 1997).