In his latest anti-cannabis-legalization screed, (behind the Wall Street Journal paywall), written with a former federal prosecutor named Robert White, William Bennett writes:
Mark A.R. Kleiman, a professor of public policy at the the university of California, Los Angeles, has estimated that legalization can be expected to increase marijuana consumption by four to six times. Today’s 2.7 million marijuana dependents (addicts) would thus expand to as many as 16.2 million with nationwide legalization.
Now, if Bennett wants to make silly predictions, and if Rupert Murdoch wants to publish them, all I can say is, “It’s a free country.” But I think I’m entitled to protest when he attributes that silliness to me. It’s hard to count how many ways that short paragraph is wrong, but the central points are simple:
1. An estimate of the possible change in quantity consumed is not an estimate of the change in the number of dependent users. Consumption can also grow because the amount consumed per dependent user increases.
2. Even most dependent users are not, by any reasonable definition, “addicts.”
3. The large estimated impact on consumption depends the  factor-of-ten price decrease (to about $1-2/gm. for moderately potent product) that would result if cannabis were treated like an ordinary commodity. If taxation or production limits prevent such a drastic decrease, the effect of legalization on consumption would be much smaller.