On October 24, 2012, U.S. District Judge Jed Rakoff sentenced Rajat Gupta to 24 months after he was found guilty by a jury of one count of conspiracy and three counts of substantive securities fraud, in connection with providing material non-public information to convicted inside trader Raj Rajratnam. This two-year prison sentence was substantially below the applicable advisory range under the United States Sentencing Guidelines and, in the week since that ruling, much has been said about whether or not this sentence was appropriate.

But the most remarkable part of Judge Rakoff’s sentencing ruling was his unflinching analysis of the way in which the application of the Sentencing Guidelines to white collar fraud cases does not reflect empirical analysis about those offenses or those who commit them – an argument that defense counsel have been making for some time with mixed success.

Judge Rakoff began his analysis with an eloquent and incisive observation about his role as a sentencing judge and the inadequacy of the sentencing guidelines as a comprehensive tool to determine a defendant’s sentence:

Imposing a sentence on a fellow human being is a formidable responsibility. It requires a court to consider, with great care and sensitivity, a large complex of facts and factors. The notion that this complicated analysis, and moral responsibility, can be reduced to the mechanical adding-up of a small set of numbers artificially assigned to a few arbitrarily-selected variables wars with common sense. Whereas apples and oranges may have but a few salient qualities, human beings in their interactions with society are too complicated to be treated like commodities, and the attempt to do so can only lead to bizarre results.

Judge Rakoff noted that the Sentencing Guidelines were “originally designed to moderate unwarranted disparities in federal sentencing” on the theory that the Guidelines “would cause federal judges to impose for any given crime a sentence approximately equal to what empirical data showed was the average sentence previously imposed by federal judges for that crime.” Of course, as the Supreme Court has already observed, the Guidelines deviated from this goal almost from the start.

For example, based on “limited and faulty data,” the Sentencing Commission determined that an ounce of crack cocaine should be treated as the equivalent of 100 ounces of powder cocaine for sentencing purpose, even though the two substances were chemically almost identical and, as later studies showed, very similar in their effects. The result of this empirically unsupportable conclusion was an indefensible racial disparity in narcotics sentencing. Kimbrough v. United States, 552 U.S. 85, 96-98 (2007). Judge Rakoff noted that, even when the Sentencing Commission changed the ratio from 100-to-1 to 18-to-1 in 2010, that ratio was likewise not based on empirical evidence but was merely “plucked from thin air.”

Judge Rakoff went on to observe that the Guidelines applicable to white collar fraud likewise “appear to be more the product of speculation, whim, or abstract number-crunching than of any rigorous methodology,” and that this “maximize[es] the risk of injustice.” Noting the huge increases in the recommended Guidelines for fraud cases, Judge Rakoff noted that the resulting advisory ranges “are no longer tied to the mean of what federal judges had previously imposed for such crimes.” Rather, these sentences “instead reflect an ever more draconian approach to white collar crime, unsupported by any empirical data.”

In short, congressional mandates to get tougher on fraud have resulted in a singular focus on one factor – the amount of loss – that “effectively ignored the statutory requirement that federal sentencing take many factors into account, see 18 U.S.C. § 3553(a), and by contrast, effectively guaranteed that many such sentences would be irrational on their face.” The result, Judge Rakoff observed, was “to create, in the name of promoting uniformity, a sentencing disparity of the most unreasonable kind.”

Regardless of whether or not one agrees with the sentence ordered in the Gupta case, Judge Rakoff’s analysis of the way in which the Sentencing Guidelines fail to promote justice in white collar cases is sure to have significant weight in other cases going forward. As structured, federal sentencing begins with a calculation of the advisory Guidelines range, and then defendants seek a variance from that range under Section 3553(a) – a process that creates a de facto presumption that a defendant will be sentenced within the Guidelines range. A recognition that the Guidelines ranges applicable to fraud crimes are not fair is a good first step towards reforming sentencing in such cases in the interest of true justice.