"
The full methodology isn't actually published anywhere."
I believe it is: http://www.thelancet.com/journals/lancet/article/PIIS0140-67.... Is a panel of experts providing views on an objective criteria the most satisfying solution? No. But they are open about the method, the scoring system, and data used to set assumptions. Plus, it beats "experts" legislating based on subjective, opaque critera.
"The rankings are creating by combining a lot of different factors that don't have anything to do with each other, e.g. by combining harm to the user with harm to society"
Nutt et all advocate considering the harm done to society by drugs, versus individuals, since it is at the society-wide level that we regulate. In any case, the data are provided to allow for one to pry apart the local and global harms. Here is a chart allowing you to, visually, do just that: http://www.economist.com/blogs/dailychart/2010/11/drugs_caus....
"The harms for drugs are measured as they are typically used, rather than correcting for things like differences in demographic and route of administration."
In his book, he spends a good deal of time addressing routes of administration, pointing out that "the health impacts of a drug can vary enormously depending on how it’s consumed". He is similarly expansive on the role of "setting" and culture in vulnerability and harm potential.
"The harms of the drugs caused by prohibition are not accounted for."
I thought he went quite extensively into the War on Drugs and how it wasn't "obviously the wrong thing to try in the 1970s, but today it is clearly doing more harm than good", that the "'drugs problem' needs radical rethinking as a public-health crisis rather than a moral crusade." He also uses the example of Portugal to show how legalisation can reduce indirect drug harms.
"They're not accounting for the benefits of drug use, only the harms."
Nutt is presently at the forefront of the British effort to de-regulate drugs. Further, harm reduction should first focus on clarifying risks.