Macleod & Hickman's story of how the United Kingdom re-classified and then re-reclassified cannabis (in 2002 and 2008, respectively) [1] illustrates neatly at least two of the Implicit Rules of Evidence-Based Drug Policy, at least as Peter Reuter and I articulated them recently [2]:

1 Evidence that a drug impairs human capacities is always believable and important.

2 Our best estimate of a drug's harm is not the average estimate but the most severe estimate yet obtained.

3 Evidence that an illicit drug could have benefits may not be collected.

4 Treatment requires evidence of both effectiveness and cost-effectiveness.

5 Evidence regarding prevention is always welcome, but it still will not gain much funding.

6 Law enforcement and interdiction require no evidence at all; they are assumed to be effective and appropriate.

7 Evidence against enforcement creates a presumption that the researcher is a liberal.

8 Evidence for harm reduction creates a presumption the researcher approves of drug use.

The story also suggests at least one new rule:

9 Scientific research on drugs cannot motivate a change from tough law to lenient law, but it can motivate a change in the opposite direction.

Perhaps one upside of this UK turnabout is that we can finally learn whether re-scheduling cannabis itself is enough to ‘send the wrong message’, encouraging more use by connoting a lack of firm resolve. The available data cannot support a rigorous econometric analysis, but for what it is worth, past-month prevalence among 16–59-year-olds in England and Wales was 6.4% before in 2000 and 6.5% in 2003–04. (It is too soon to say whether the 2008 re-rescheduling mattered.) In comparison, 30-day prevalence stayed flat in Germany, Spain and Italy and rose slightly in Finland and Sweden during this period [3].

After Macleod & Hickman carefully document the uncertainties surrounding the cannabis–schizophrenia association, they then offer a sweepingly casual assessment of drug policy:

The most rational policy on cannabis from a public health perspective would seem to be one able to achieve the benefit of reduced use in the population while minimizing social and other costs of the policy itself. Prohibition, whatever the sentence tariff associated with it, seems unlikely to fulfil these criteria.

I very much agree with the first sentence, but the second one seems awfully brash. If rigorous causal identification is difficult for the cannabis–schizophrenia link, it is surely nearly impossible for the prohibition–costs link (or links)—we lack even a rough case–control study of the effects of prohibition, much less the controlled trial one would really like to establish the efficacy of the intervention. There are good reasons to argue that drug prohibition is a mess; we just do not yet know for sure whether the alternatives would work better.

It is tempting to counter by arguing that realistic policy analysis requires a more lax view of evidence; decisions have to be made and the failure to act is itself costly. However, if we take this stance, we risk losing any basis for contending that the 2008 UK re-rescheduling involved ‘jumping to premature conclusions’—a sentiment of Macleod & Hickman that I find myself sharing.

Perhaps this suggests one more rule [4]:

10 Experts like to have it both ways; we hold the government to higher standards of proof than we apply to our own policy opinions.

Declaration of interests