NPR on the Crack Cocaine Sentencing Rules:

NPR ran a story Friday on the 20th anniversary of the crack cocaine sentencing rules — rules under which the weight thresholds for crimes involving crack cocaine are set 100 times lower than the thresholds for crimes involving powder cocaine (so that dealing 5 grams of crack will be treated comparably to dealing 500 grams of powder). There surely are important criticisms of these rules; I'll implicitly touch on one below.

But, boy, the NPR story seemed weak, in pretty obvious ways. I'm not a drug policy expert, which is why I almost never blog about the subject, but the story's weaknesses are so clear that it seems to me one needn't be an expert to spot them. A few examples:

The U.S. Sentencing Commission has recommended reducing the differential between crack and powder cocaine since 1995.

It's found that 83 percent of the people in prison under the law for crack cocaine are African-Americans, even though many users of crack cocaine are white.

Let's begin with the comparison — 83% of the people in prison for crack cocaine are black, though many users are white. How many? "Many" could mean 5% of a large number, or 20% of a large number, or 60%. I've certainly heard that the actual percentage is enough to make for a substantial disparity between the demographics of imprisoned crack criminals and the demographics of users. But the story doesn't even assert that, much less that demonstrate that with numbers.

Second, it's far from clear that there's anything sinister in treating drug dealers more harshly than drug users; the premise of most controlled substances laws has been precisely that. (Even critics of drug laws rarely argue that the solution is to impose the same long sentences on users as on dealers.) If the Mafia were running drugs to predominantly black neighborhoods, you'd hope that the great majority of drug prisoners were of Italian extraction even if the great majority of users were black. Even if drug users were punished to some extent, they'd probably be sentenced to shorter terms, which would lead to the disproportion between the demographics of prisoners and of users. Likewise if crack cocaine distribution is mostly a black business and crack cocaine buyers are mostly white.

Now if you think racial disproportions matter, you might compare crack dealers to dealers in other drugs, and see whether black drug dealers were getting much higher sentences than white drug dealers (controlling for the harmfulness of the drugs). I can't tell you what those numbers would show. But those aren't the numbers that the NPR story gave, or, more precisely, hinted at (given its 83% vs. "many" comparison).

Here's the next sentence:

And it's found that most of those imprisoned are street dealers or drug couriers, not kingpins.

Most of those imprisoned aren't kingpins? So what? Under any sensible definition of kingpins, kingpins would be only a tiny fraction of the distribution pyramid. Even if the war on drugs is perfectly proper, and even if it is conducted in the most efficient and racially evenhanded way, and even if law enforcement tries much harder to get kingpins than it does to get the lower-level, medium-level, and the sub-kingpin-high-level participants, of course kingpins would constitute only a small minority of those imprisoned. (If you have a hard time imagining such a scenario, because you so reject the war on drugs, imagine an attempt to fight only those who are distributing highly addictive drugs to minors, or an attempt to fight extortion rackets.)

The only way you can get a situation where "most of those imprisoned are kingpins, not street dealers or drug couriers" is if you virtually eliminate any prosecution of street dealers or drug couriers. That's hardly a sensible strategy of fighting the war on drugs (as opposed to of abandoning it). Again, maybe there's a serious criticism hiding behind the vague assertions, for instance that the rules are highly ineffective against kingpins (or that no matter how many kingpins you lock up, some others will rise to take their place). But if that's what you're trying to say, say it. Don't give a vague assertion that, even if completely correct, would tell us nothing about whether the law is working well.

Then the story goes on:

[M]ore voices are calling for change. Senator Jeff Sessions, the conservative Republican from Alabama, is one of them. He's introduced a bill that would reduce the differential between crack and powder cocaine from a hundred to one; to 20:1.

Senator JEFF SESSIONS (Republican, Alabama): And now we have had nearly 20 years of experience and I think, legitimately, based on my experience as a federal prosecutor, that the crack sentencing guidelines are too heavy. And there's — it's not necessary to have as long of sentences for some of these offenses as we now have. And it's appropriate if Congress is going to move in to this area, that it review what it's done and — and make adjustments as time goes by.

OK, so 100:1 is supposedly bad — but how are we to decide whether the 20:1 would be any better? Why not 5:1? Why not 1000:1? Are we just supposed to figure out the ratio at which the racial demographics of prisoners and users match (the main criticism of the 100:1 disparity that the story had pointed to so far)? Or are we just supposed to figure out the ratio at which the majority of prisoners will be kingpins, in which case why would a ratio help at all?

Now as I've mentioned above, there are some sensible factors one can consider, if both crack and powder cocaine are to be outlawed, and we're looking for the right weight equivalence ratio. Most important of them, it seems to me, would be how dangerous crack is per gram compared to cocaine.

The theory for calibrating punishment to weight more generally, even within the same drug, is that 1000 grams of cocaine are thought to be more dangerous than 10 grams, because they represent more doses. (Again, if you think that neither is dangerous enough to justify the war on drugs, think of what policy you'd use for deciding how to punish people who were trying to sell cocaine to children.) One reasonable theory for calibrating punishment to drug type would be if some drugs were thought to be more dangerous per dose than others. If one gram of crack yielded approximately as much harm as 100 grams of cocaine, whether because of difference in dose size or dose harm (recognizing of course that any such matters are only estimates), then the 100:1 disparity may well be sensible. If our best guess of the harm ratio is that 1 gram of crack yielded as much harm as 1 gram of cocaine, then, unless there are some other factors that the harm analysis doesn't take into account, there shouldn't be a disparity at all. If the harm-per-gram ratio is 20:1, then we might want to consider a 20:1 weight ratio.

But even if I'm wrong in this analysis, at least it's an example of how one can think of the matter in a way geared to actually reach a sensible policy result, rather than just reporting on a proposed number with no explanation for it, or quoting people who would replace one number with another, with no explanation for the reasoning that might support either. Some such sensible analysis, whether the one I offer or a better one, would be more helpful to listeners than "lots of people say 100:1 is too much, and it leads to racial disparities of a magnitude that we'll only hint at, so let's try 20:1."

UPDATE: In the last paragraph, I originally wrote "picking numbers out of thin air" instead of "reporting on a proposed number with no explanation for it" and "who would replace one thin-air number with another" instead of "who would replace one number with another, with no explanation for the reasoning that might support either"; this post by Doug Berman reminded me that my formulation wasn't quite right -- I meant to fault NPR's reporting for giving numbers with no explanation, but my "thin air" reference erroneously suggested that the numbers had no explanation at all. I can't speak to that latter question; for all I know, for instance, Sen. Sessions' 20:1 proposal is eminently sensible and well-supported -- my point is simply that NPR gave us nothing to explain the numbers, which for all we were told appeared out of thin air. My apologies for the error, and my thanks to Prof. Berman for prompting me to correct this.