Ezell Gilbert is now before us asking to be relieved of the consequences of a mistake we made in his direct appeal in 1998. He told us then that the District Court was wrong in sentencing him substantially more harshly based on that court’s decision that carrying a concealed weapon is a crime of violence. We rejected his argument, and affirmed his sentence of more than 24 years. United States v. Gilbert, 138 F.3d 1371 (11th Cir. 1998). We did this on a record containing the District Judge’s clear statement that the sentence was longer than he would have imposed, but for the then-mandatory Sentencing Guidelines. Id. at 1372–73. It turns out, of course, that Mr. Gilbert was right and we were wrong. Carrying a concealed weapon is not a crime of violence. We said so, belatedly for Mr. Gilbert, in United States v Archer, 531 F.3d 1347 (11th Cir. 2008).
The effects of our mistake are quite dire for Mr. Gilbert, insofar as his
properly calculated (and advisory) guideline range would today be 130–162
months, or approximately 11 to 13 years. As I write this, I understand that he has already served more than fourteen years in prison. And yet the majority opinion tells Mr. Gilbert that the laws and Constitution of this country offer him no relief.
From the dissenting opinion of Judge Martin from an en banc ruling of the 11th Circuit in an appeal of a U.S. District Court denial of a habeas corpus petition.
The majority opinion by Judge Carnes (at the same link) opens with the following:
Ezell Gilbert, a federal prisoner, wants to have an error of law in the calculation of his sentence corrected based upon a Supreme Court decision interpreting the sentencing guidelines, even though that decision was issued eleven years after he was sentenced. Gilbert insists that prisoners have a right to have errors in the calculation of their sentences corrected no matter how long it has been since the sentences were imposed. . . . The principles of policy that limit the right to be resentenced in accord with the latest guidelines decisions are those regarding finality of judgment and the important interests that finality promotes. For reasons we will discuss, the statutory provisions and the decisions furthering finality of judgment are strong enough to hold their own against Gilbert’s claimed right to have a long-ago error in calculating his sentence corrected.
Gilbert had a prior criminal record as a drug dealer, and the sentencing was conducted on the basis of plea bargain that stipulated that "his statutory sentencing range was ten years to life imprisonment on the crack distribution count and not more than five years on the marijuana distribution count. . . . Gilbert’s base offense level was 32 because his distribution offense involved at least 50 grams but less than 150 grams of crack." His lengthy and serious criminal history produced a sentencing guideline sentence about twice what appears to have been anticipated by either party because one of his prior convictions for unlawfully carrying a concealed weapon was treated as a violent rather than a non-violent offense.
The majority makes much of the fact that Gilbert received many instances of leniency in his prior run ins with the criminal justice system and could have faced a mandatory life in prison sentence if the prosecutors had chosen to take a fairly strong case to trial rather than plea bargaining. Basically, the majority argues rhetorically, although not really legally, that bad people like Gilbert deserved to suffer from judicial mistakes anyway, and gets to benefit from mistakes made in his favor, so judicial mistakes of law aren't worth correcting when they prejudice him.
In contrast, as Gilbert's lawyers see it, the basic problem is that he got a sentence twice as long as the one that he had bargained for (and of course, it bears noting that this long sentence was only made possible by laws imposing hysterically long sentence for dealing crack cocaine in small quantities that the Congress has since disavowed.)
The 11th Circuit is one of the most conservative of the U.S. Court of Appeal Circuits in the United States federal court system. The case eptiomizes the divide between liberals and conservatives over habeas corpus relief. (Note that this was a federal court decision, so the federalism aspects of habeas corpus law are absent from this case.)
It also illustrates the fact that in the federal sentencing guidelines system, slight nuances of interpretation regarding gray areas of the governing law can have profound consequences for criminal defendants. This suggests that there are real problems with the fundamental structure of this criminal sentencing system, which in practice can be quite capricious.
Then again, there is wide bipartisan consensus that far more profound mistakes in fact finding by juries cannot be corrected on any kind of appeal, and these mistakes are probably considerably more common than mistakes on questions of law that are subject to appeal or correction in a collateral attack on a conviction.
Reasonable estimates put the error rate of juries on the question of conviction for a crime, based on comparing the conclusions drawn by juries and judges and other judges with each other based on the same evidence and based on exoneration rates in cases where DNA evidence is available, at 10%-20%. Of course, something on the order of 90%+ of cases are plea bargained, so jury errors make up only about 1%-2% of criminal prosecutions. But, this still probably dwarfs the number of material mistakes made by judges on questions of law that are not corrected on direct appeal. While fixing mistakes that are clearly wrong, even if discovered and accurately addressed only much later, makes the system more accurate, errors in fact finding and poor legislative and quasi-legislative standards for sentencing across the board are a bigger problem in the inaccuracies of the criminal justice system as a whole.