In the first set of experiments below we do this with a simple threshold, relying on the ability of an expert designer to apply experience, or some established rule of thumb, to give an upper bound on the total weight of plausible solutions. This upper bound does not need to be a tight upper bound, and in fact a bound that is too tight might prevent the GA from converging at all. We only assume that a bound that is within a factor of five to ten of a desirable solution quality can be given. In the second set of experiments we look at an algorithm that sets the threshold dynamically.