Is majority opinion a better guide to action than a minority view? This paper demonstrates that a direct application of rational herding theory to this novel area can produce a surprisingly counter-intuitive result: given (i) the minority has a clear conformist view and (ii) decision-makers learn through observation as in a herding model, then size does not matter when evaluating whether some groups make better decisions than others. Extending this further we argue that it may be advantageous for risk averse agents to support a form of positive discrimination, that new generations have a largely ambiguous impact, and that the use of electoral colleges can be supported on informational grounds.