David Frum and I agree that “But something else is even worse than X!” is not a good reason to ignore the X problem: autos kill more people than guns, but we should still try to reduce the number of people killed with guns. And the fact that alcohol is a much nastier drug than cannabis, both physiologically and behaviorally, doesn’t make cannabis abuse either rare or benign.
But Point #13 in the post Frum links to wasn’t about the comparison between cannabis and alcohol; it was about the causal connection between cannabis policy and alcohol abuse. As Frum notes, alcohol use and cannabis use are now positively correlated. But that doesn’t tell you anything conclusive about whether making cannabis legally available would increase or decrease heavy drinking.
In my view, an increase of as little as 10% in heavy drinking would wipe out any benefits from cannabis legalization, including the benefit in the form of fewer arrests because of the additional crime that would go along with the additional heavy drinking. Frum is aware of that possibility.
But he ignores the opposite possibility, equally plausible in terms of both logic and evidence. If legalizing cannabis (under some specified set of taxes and regulations, including, for example, a ban on lacing beer with cannabinoids) turned out to decrease heavy drinking by 10%, then any “public health case against cannabis legalization” would vanish in – pardon me – a puff of smoke.
Since the benefit-cost analysis of cannabis legalization turns crucially on its effect on heavy drinking, and since that effect is unknown, dogmatic assertions about whether legalization would, on balance, be beneficial or harmful are not justified by the current state of knowledge. (Principled support for legalization on libertarian grounds, or principled opposition to it on cultural-conservative grounds, remain logical possibilities.)
The question seems to me close enough, and the risks limited enough, to justify an experimental approach: letting Colorado and Washington go ahead with state-level legalization to see what happens. Someone else, assigning different probabilities to the possible outcomes or different weights to the categories of gain and loss, could reasonably oppose even experimental legalization as a bad gamble.
But someone who says “I know that the harms of legalization would outweigh the harms from prohibition” (or vice versa) is, it seems to me, fooling himself, allowing his cultural preferences to influence his judgment about empirical matters in just the way Dan Kahan describes.