[davidw]

I think the more you add context and statistics, the more you find a variety of methods and outcomes, which is why you get the equal and opposite bits of advice; no one thing works in all cases.