Half&half bagging is a method for generating an ensemble of classifiers and combining them that does not resemble any method proposed to date. It is simple and intuitive in concept and its accuracy is very competitive with Adaboost. Certain instances that are used repeatedly turn out to be located in the boundaries between classes and we refer to these as hard boundary points. The effectiveness of half&half bagging leads to the conjecture that the accuracy of any combination method is based on its ability to locate the hard boundary points.