Modelling research and development: How do firms solve design problems?

One way of thinking about research and development is to recognise that firms are trying to solve particular design problems. We often build these design problems into our models, but are forced to oversimplify them in order to make the models solvable. The approach taken in this paper is to acknowledge that design problems are often insoluble using standard techniques and to model instead the process by which firms solve them. Two such processes are simulated in detail. The first, individual experimental search, is based on a problem-solving technique known as simulated annealing. The second, partial imitation, involves learning at a social level and is based on a problem-solving technique known as the genetic algorithm. Some economic implications of these processes are explored, including their application to stochastic learning curves, patent design and the importance of `technodiversity' in the introduction of new technology to developing countries.