I think Taleb's argument is more along two lines:

1. Hormesis ("low-dose exposures have a beneficial effect and high-dose amounts are either inhibitory to function or toxic"[0])

2. Evolutionary adaptation: alcohol has been around humans since literally forever, it has even been a part of our evolutionary ancestors' diet for millions of years. For us, it occurs in (over)ripe fruit, bread, and in some of us via natural fermentation in the stomach. Directed creation of alcoholic beverages has been done for more than 10.000 years, on top of that. This puts alcohol in a completely different risk class of substances compared with other legal or illegal drugs, because risk is a function of (mal)adaptation. I'd be inclined to limit this argument to fermented products and exclude distilled spirits.

3. "Drunk hypothesis": the true beneficial effect on humans could be failed to be measured by life expectancy of a single human, and instead should be measured by the "life expectancy" of societies and their cultural dominance as [1] hypothesizes.

[0] https://en.wikipedia.org/wiki/Hormesis

[1] https://www.mpg.de/25055517/alcohol-and-the-rise-of-complex-...