Alcohol is toxic to most major organs, is one of only a tiny number of drugs that can make normally passive people become violent, and is the world's most prolific date-rape drug. If you've ever seen the emergency room in a typical hospital on a Saturday night, you'd see that many of the millions who enjoy a drink get violent and antisocial, injuring themselves and others. Then there's domestic abuse, drunk driving, risky sex, etc etc.
You'd need to take figures on all of those effects into account before deciding whether it's a 'stretch' or not. Which is exactly what Prof. Nutt did in his paper.
I do not believe that alcohol is a net negative to society. I'm pretty confident that crack cocaine is more damaging, especially to the individual user (and in fact you'll find that borne out by Nutt's research).
However, while perhaps "nobody robs a granny to buy a single malt," alcohol can very obviously have damaging effects, especially if it's misused. Sexual assault and drunk driving are just two examples.
That these exist does not mean that alcohol should be banned, and that crack is more damaging does not mean that the most effective tool to prevent it's use is criminalisation.
No, they rob granny to buy another bottle of $15 vodka.