save social security?

http://freakonomics.blogs.nytimes.com/2008/09/19/can-binge-drinking-save-social-security/

The funny thing is that this isn't a new concept. Smoking has the same (well, similar) effect on general mortality as heavy alcoholic consumption. There are two flaws with the assumptions: 1) Social security will remain as it is. I highly doubt that. And 2) heavy drinkers are essentially distributed throughout the general population at random, which studies have shown is usually not the case.

In fact heavy drinkers are generally distributed in mostly lower income jobs after college (not that rich people can't be alcoholics, but rather that they are statistically of a different sort than beer drinkers). Rather than redistributing 'wealth' from heavy drinkers to non-heavies, it basically redistributes scraps. Drinking in general, as has been known, is not by itself a cause of un-health, rather its extreme is. And this extreme is basically taken into effect by people who are too poor to deal with the effects (often referred to as 'college students'). In other words, even if they generate several billion dollars of extra social security monies that they won't live to see, wouldn't they also generate several billion dollars of other social costs; such as health care, lost productivity, accidental damages, and insurance claims? Wouldn't these off-set the positive externality of people dying off too fast to get their money?