A recent First Circuit decision raises three important sentencing issues. While framed as a decision about the right of defendants to learn the identities of confidential informants, last week’s United States v. Mills opinion impacts not only that issue, but also the fairness of the United States Sentencing Guidelines’ penchant for driving up sentences dramatically based on uncharged conduct, and the use of “reasonable estimates” as the bases for sentence enhancements.
Two days before the First Circuit’s Mills decision, Judge Jed Rakoff of the Southern District of New York, an outspoken critic of the United States Sentencing Guidelines, urged the “scrapping” of the Guidelines’ use of formulaic upward adjustments of sentencing ranges based on the amount of loss a court calculates from the crime. Judge Rakoff has previously written of “the utter travesty of justice that sometimes results from the guidelines’ fetish with abstract arithmetic, as well as the harm that guideline calculations can visit on human beings if not cabined by common sense.” United States v. Adelson, 441 F. Supp. 2d 506, 512 (S.D.N.Y. 2006).
Just as Judge Rakoff was lamenting the arithmetical absurdity that the Guidelines produce, the First Circuit in Mills affirmed a sentence of 108 months’ imprisonment based on upward adjustments under the Guidelines that increased the sentence from just 24 months – based entirely on conduct for which the defendant was never even charged, let alone convicted.
The First Circuit was addressing calculations of drug quantities, which function similarly to the loss calculations in white-collar cases that Judge Rakoff criticized, in that both produce skyrocketing Guidelines ranges as the numbers grow. Whether loss amounts or drug quantities, the numbers used for Guidelines purposes are calculated by a judge, not a jury; can be based on mere “reasonable estimates”; can be “proven” through evidence that would not be admissible at trial, including hearsay; and are not subject to the proof-beyond-a-reasonable doubt standard. In Mills, the defendant was caught smuggling 104 eighty-milligram and five forty-milligram oxycodone pills into the country from Canada. The defendant pled guilty to the one count with which he was charged, without a plea agreement. At sentencing, the court imposed an enhancement to the Guidelines calculation based not on the 109 pills with which the defendant was caught and for which he was charged, but rather based on 2,637 eighty-milligram pills. This enormous increase resulted from uncharged conduct – prior smuggling trips that the defendant was theorized to have made, but for which he was not convicted, or even charged.
Worse still, the government supported its request for a sentence based on this uncharged conduct with information from three confidential informants. It did so while refusing to identify the informants to the defense. The defendant argued to the First Circuit that the increase from 109 eighty- and forty-milligram pills to 2,637 eighty-milligram pills added seven unexpected years to his sentence.
In one sense, Mills is routine in allowing the government to rely on hearsay and a sentencing court to make an “estimate” that drives up a sentence. Here, however, the resulting sentence that was 450% of what the defendant expected before the effect of the “estimate” essentially convicted him of multiple offenses instead of just the one to which he pleaded guilty, without the many safeguards that would have accompanied an effort by the government to actually convict him of the other conduct.
The Mills court’s analysis, while giving prosecutors extensive room to use untested and untestable “evidence” to support huge increases in sentences, nonetheless provides some guidance for what defendants and defense counsel can do or should not to do in sentencings in which an extreme “estimate” is challenged and the identities of confidential informants are sought.
First, Mills was critical of the defense’s reliance on what the court viewed primarily as speculation about the questions it would ask and the ways in which it would impeach the informants. Ironically, the sentencing judge rejected the defense’s alternative explanation for large sums of money (which the defense said came from money-laundering activities, unrelated to oxycodone smuggling) because the defense refused to identify the person for whom the money was being laundered, “leaving no way to verify his claim.” This reasoning seems inconsistent with allowing the government to hide the identities of informants, other than on a theory that the government is more trustworthy. Putting aside the inherent challenge a defendant faces in specifying the line of attack he will use when he is not permitted to even know the identities of the witnesses against him, nothing in the opinion suggests that the defendant attempted to ask the agents who drafted the memoranda of interview questions about the reliability of the unnamed witnesses.
Second, the defendant had posted portions of the informants’ reports on Facebook, and others involved with the defendant were found carrying documents from the case, which arguably threatened disclosure of the witnesses as “snitches.” In light of these facts, the First Circuit appears to have given significant weight to the serious consequences that might befall the government’s informants (although says little about the serious consequences – years more in prison – that the defendant faced). There appears to have been no request for, or consideration of, a protective order that would allow the defense, or perhaps even defense counsel alone in light of the Facebook posts, to learn the witnesses’ identities without the identities being made public.
Third, the opinion relies on an implicit assumption that government agents are to be trusted merely because their statements come with the imprimatur of the government – an assumption that defense counsel must be willing to contest. While describing the informants’ statements as more akin to corroborating evidence than essential evidence, the opinion repeatedly relied on the statements and repeatedly noted that they were generally consistent with one another and with the government’s theories. As it was government agents who were presenting that hearsay evidence as consistent with the government’s theories, one might reasonably question the reliability of the reported consistency – exactly the type of concern that cross examination and hearsay rules are designed to address.
Finally, absent from the opinion is any discussion of an argument that more reliable evidence should be required to support the purported “reasonable estimate” used for Guidelines purposes when the result is both a de facto conviction of multiple uncharged crimes and a sentence that is a multiple of what it would likely have been in the absence of those other crimes for which the defendant was neither charged nor convicted. What might be reasonable to support a modest increase should not necessarily be deemed reasonable for a dramatic one. Defendants should insist upon a form of sliding-scale burden, meaning the need for a higher and better-quality evidentiary basis for an estimate being deemed “reasonable” where that estimate has such dramatic effects on a sentence.
Mills is perhaps unremarkable under the state of the law in the First Circuit. But the combination of the government’s ability to rely on sources it refuses to identify – and other evidence that it could not have admitted at a trial – the Guidelines’ acceptance of “reasonable estimates” drawn from such inherently unreliable evidence, and the harsh consequences that flow from mathematical Guidelines calculations linked to such evidence and estimates, should undermine faith we might have in the fairness of the Guidelines-driven sentencing process and should support calls such as Judge Rakoff’s that they be “scrapped.” But while they remain in place, defense counsel should use the room that the reasoning of the First Circuit’s opinion and similar decisions provide to bust the “reasonable estimates” that the government offers and the “tattlers privilege” that it uses to conceal its evidence.