the direct biological harms of alcohol are roughly linear, i.e. one drink a week is about 1⁄7 as harmful as seven drinks a week, which in turn is about 1⁄7 as harmful as forty-nine drinks a week;
the psychologically mediated benefits of alcohol (through reduced anxiety, improved socialization, etc.) quickly rise up to a few drinks a day, then plateau (and even reverse at very high doses)
when you subtract something like atan(x) from a straight line you may or may not get a minimum at x slightly greater than zero, depending on the slopes involved
My guess would be that:
the direct biological harms of alcohol are roughly linear, i.e. one drink a week is about 1⁄7 as harmful as seven drinks a week, which in turn is about 1⁄7 as harmful as forty-nine drinks a week;
the psychologically mediated benefits of alcohol (through reduced anxiety, improved socialization, etc.) quickly rise up to a few drinks a day, then plateau (and even reverse at very high doses)
when you subtract something like atan(x) from a straight line you may or may not get a minimum at x slightly greater than zero, depending on the slopes involved