I agree that there is good evidence that people who don't drink die more often and/or have more heart disease. My speculation is just that it might not be a physiological effect of alcohol consumption that's causing this. Higher levels of stress and lower levels of social interaction are both correlated with all sorts of negative health effects, and I think it's plausible that moderate drinkers are just generally more social and less stressed people than people who abstain.
I shouldn't have said that the research is flawed, because it's (probably) perfectly good observational research that has established a correlation between moderating drinking and increased longevity. I'm just not sure it supports the conclusion that drinking, in the absence of any other changes, will improve someone's health.