Red wine is one of those health subjects where the pendulum seems to swing back and forth a lot. Good for you in moderation, bad for you in excess. No bad for you even in moderation, etc.
But where research says "Red wine = good" the causal factor for that 'goodness' is the molecule shown at right. This is the molecule resveratol. On this point it seems to me most researchers are agreed.
Consider me a skeptic on this point. Red wine greases social interaction through the disinhibitory effects of alcohol and is steeped in culture and history. If there is an associated positive health effect of red wine, perhaps it is not simply though the effects of its parts (resveratol) but in its role in more fundamental wholes. Any good health researcher is aware, of course, that there may be unknown variables that may be the overlooked active ‘ingredient’. In the jargon of epidemiology, these hidden factors go by names such as confounders and colliders and, when unrecognized, may lead to faulty inferences. Despite this recognition, both health research and social science utilize methods borrowed from the natural sciences that are themselves predicated on this mechanical worldview that looks for the answer in the constituting parts. Thus, in the progression of a typical health study, researchers observe a particular outcome in a group they are following – such as a positive benefit of red wine. They then isolate the substance from its natural context, honing in on the active ingredient through an experimental research design aimed at reducing confounding variables and noise. If it's effect is being embedded in a larger social activity that is really critical, then such a reductive approach to learning at best will fail to find an effect. At worst, the many biases and false positives in health and social science research will creep in, leading investigators to affirm faulty hypotheses and further a reductionist world view.