12% of people in the US have had alcoholism according to Wikipedia. That is a minority of people. The trick is to distinguish those at risk and advise zero consumption to them AND THEM ONLY while advising moderate alcohol consumption to the majority of people for whom such consumption would yield optimal outcomes along many axes from better social cohesion and lower mortality. That is for the population as a whole, with some exceptions for mentally ill people but not in a blanket prohibitionist way.

There is no controversy that the relationship between alcohol consumption and positive health outcomes is not linear. It is not the case that zero consumption yields the best outcomes and the more a person consumed, the worse the outcomes become, progressively. Nope, the curve has a U-shape; in other words, there exists a sweet spot of optimal moderate alcohol consumption that outperforms BOTH abstinence and heavy consumption. The disagreement is only about what exactly the sweet spot is - 1 serving a day? 2 servings a day? 1-6 servings a week? etc. There is no disagreement regarding the shape of the curve.

Advising zero consumption, assuming that people follow the advice, optimizes the outcomes for the 12% of the high risk population while treating the low risk 88% as collateral damage, denying them the many benefits that would accrue to them should they be advised to consume alcohol in moderation.

To advise zero consumption is akin to performing a C-section on every pregnant woman. Sure, this way we would definitely catch those few high risk women that really need a C-section, but at the cost of worse outcomes for the majority of women and their babies.

It is also akin to the now outdated practice of prescribing antibiotics indiscriminately. Sure, if a doctor prescribes antibiotics to everyone who enters his office, he will most definitely prescribe them to those few who really need them.

In information retrieval terms, the zero consumption advice has only one benefit - of perfect recall, with respect to alcoholics. But the precision is absolutely horrendous.

Usually, one tries to find a way to balance precision and recall.

More on them:
http://en.wikipedia.org/wiki/Precision_and_recall

In pattern recognition and information retrieval, precision is the fraction of retrieved instances that are relevant, while recall is the fraction of relevant instances that are retrieved. Both precision and recall are therefore based on an understanding and measure of relevance. Suppose a program for recognizing dogs in scenes identifies 7 dogs in a scene containing 9 dogs and some cats. If 4 of the identifications are correct, but 3 are actually cats, the program's precision is 4/7 while its recall is 4/9. When a search engine returns 30 pages only 20 of which were relevant while failing to return 40 additional relevant pages, its precision is 20/30 = 2/3 while its recall is 20/60 = 1/3.

So blanket zero consumption advice correctly catches ALL alcoholics, who are in the minority. Hence, it has perfect recall.

A better policy is to distinguish between at risk people and advise zero consumption to them and them ONLY, while advising moderate consumption to the low risk population.

There are some changes when you apply it to mentally ill people due to individual reactions and individual drug interactions, but the main point remains is that the zero consumption advice given to 100% of people is good only in that it serves the whatever percentage of dual diagnosis (with alcoholism) folks is AND NOBODY ELSE.