[spangry]

Indeed. In the context of a publicly-funded healthcare system (which Australia has), the direct health costs of alcohol consumption do end up being 'socialised'. While my (minority) view is that this argument doesn't apply to smoking, I think it does for alcohol.

Although, it is much more tricky to estimate the harm caused by alcohol consumption and then levy taxes accordingly. The reason being that unlike smoking, which has very high rates of addiction, the vast majority of people who consume alcohol don't really come to any harm because of it. But 10-15 % of consumers are super badly harmed by it, mostly due to a genetic predisposition to alcoholism. It's similar to gambling in many ways.

Ideally you'd have a tax rate that increases with each additional 'unit' of alcohol purchased within some time period. Obviously this is totally impractical to implement. But before all that there's a pretty basic 'first-step' we should be taking: levying tax on alcohol on a volumetric basis, rather than having different rates for beer, wine, spirits etc. It's the best starting point: simply levy a tax of x cents per milli-litre of alcohol. Only after we've done that can we start getting all fancy with measures like 'pre-commitment', time-varying tax rates (which I personally think are unfeasible) etc.

EDIT: As for negative externalities, putting aside public health, my guess is the two main sources would be (a) drunk, anti-social behaviour and (b) third-party road injuries and fatalities. As to the magnitude of these, I have no idea...