Just to give an idea for the algorithm, you don't need to necessarily have it be random to have it be blind. For example, you could take the costs given and multiply that cost (just for this algorithm, that is) based on priorities - a "Militaristic" research priority, for example, might multiply...