...if the line between legal and illegal drugs needs to track some imagined line between "good drugs" and "bad drugs," drug policy is profoundly incoherent. But there's no need for such a link. There are, among intoxicants, no "good" or "bad" drugs. Both alcohol and cocaine are consumers' goods with a peculiar mix of risks, including both the risks due to intoxication (loss of self-command over behavior in the short run) and addiction (chronic loss of self-command with respect to consuming the drug itself).
The question, drug-by-drug and comprehensively, is what mix of policies would minimize aggregate damage, net of benefit. (This elides the distinction between harm to self and harm to others, which strikes me as a reasonable thing to do in the face of an activity where individuals can't be assumed to be good stewards of their own well-being.) In some cases that least-cost solution will look like prohibition; in others it will look like regulation and taxation. It's a practical problem, to be handled by practical means-ends reasoning, not by the enunciation of profound truths about human nature or the role of the state.
Even believing that alcohol, on balance, creates a net social deficit, I don't actually believe that alcohol should be prohibited. Given the enormous user base for alcohol, its prohibition would be operationally nightmarish as well as politically infeasible. Instead, why not ban its sale to those previously convicted of alcohol-induced violence or repeated drunken driving? That ban wouldn't be perfectly obeyed, but it would have some good effect nonetheless, and wouldn't create another huge illicit market.
And on arguments to legalize cocaine with a very high excise tax:
The hypothetical licit cocaine market, by offering users a product of known composition, available in safe surroundings from non-criminal suppliers, and without the risk of legal sanctions at the price illicit dealers charge for product with none of those attractive qualities, would represent a substantial hedonic price decrease compared to the current situation. Moreover, legality, and the associated marketing activity, would tend to change current attitudes and opinions in a direction favorable to cocaine use.
Therefore, one would expect a fairly large increase in the number of people trying cocaine. Some proportion of them would become habituated, and some proportion of that group would find the habit difficult to break and become chronic heavy users. For them, the high legal price would mean that their cocaine habits, though legal, would be ruinous financially, and some would be likely to turn to crime to finance continued cocaine purchases. In addition, of course, they would face many, though not all, of the health and psychological risks attendant on heavy cocaine use as we now know it.
So I conclude that legalization at a high price is probably a loser, even if one adjusts for what would certainly be benefits to supplier and transit countries and some marginal reduction in resources available to terrorist organizations. That even someone as smart as Levitt could have been taken in testifies both to the peculiarities of drugs as consumers' goods and to the strength of the persistent hope that our seemingly intractable drug problems might be made to yield to some simple piece of ingenuity.
[Legalization at a low price would probably succeed in reducing overall crime, and might well be a substantial boon to poor minority neighborhoods now wracked by cocaine dealing, though at what would very probably be a very high cost in increased cocaine abuse. It is even possible, though I doubt it, that low-price legalization would represent a net improvement over today's version of prohibition.]