"Not only do I think it is meaningful to compare the harm done by drugs as they are actually used, it is necessary. Nutt's research is from society's perspective - that's why it spends more time on how people actually use drugs than how much harm would be done if they used them appropriately."
The problem is that how drugs are used is almost entirely dependent on the public policy surrounding them. E.g. when coffee was an illegal drug, you had many people abusing it similar to what we think of as a stereotypical heroin addict today. So how can you set future public policy based on the societal harms of a drug without taking into account current public policy, when those harms are mostly determined by current public policy? Especially since the study (and his marketing of it) portrays these rankings as being objective and scientific.
And even if you were to think of it as a measure of societal harm caused by individual drugs, it doesn't even make any sense for that. E.g. there are no objective measures of the number of users, the amount of harm caused per user, the amount of harm each user causes to society, etc. It's only the opinions of a bunch of random people about these things. How is this supposed to be more accurate than getting actual data?