So the observation that different drugs show different addictive potential (e.g., heroine versus cocaine) is in fact irrelevant because "addiction doesn't really have anything to do with the molecule in question"?
So I think there are three different metrics to look at:
1) How physically addictive a drug is. E.g. withdrawals from oxycontin are much worse than withdrawals from cocaine.
2) How 'moreish' the drug is. In other words, how compulsively do users redose. With salvia most people smoke it once and then don't want to do it again for at least a few years if ever, whereas with mephedrone people snort it and then they NEED to keep snorting it ever fifteen minutes or so until they run out.
3) The percentage of people that use a given drug who become addicted. You would expect this to vary wildly depending on factors one and two, but it's actually virtually a constant.[1] Pretty much 10% of people who try a drug will get addicted whether it's alcohol, marijuana, heroin, cocaine, etc. There is a little variation of course, but not nearly as much as you'd expect if addiction were driven mainly by the properties of the drug itself.
[1] Once you cancel out the differences due to the routes of administration, i.e. smoking a drug is vastly more addictive than eating it regardless of what the substance.
There's a distinction between substances that interact with brain chemistry in a particular way to enhance addictive behavior, and the circumstances that predispose some people more than others to seek out addictive substances as a means of self-medication.
So the observation that different drugs show different addictive potential (e.g., heroine versus cocaine) is in fact irrelevant because "addiction doesn't really have anything to do with the molecule in question"?