Odysseus

Background

Understanding the (social) process of technological invention is important as the invention of new technologies and technological components impacts on social welfare. New products contribute to welfare through satisfying previously unmet needs or satisfying existing needs at lower cost. Despite the importance of technological invention, little is known about the actual process of invention, as distinct from the processes of commercialisation and diffusion of inventions once uncovered. Inventions can arise from the discovery of new components and / or from the discovery of new ways to link existing components. The number of possible combinations of components is vast, and therefore invention cannot feasibly result from an enumerative or random process. Rather, inventors employ search heuristics.

Technological invention does not take place in a vacuum. It is crucially driven by inventors’ expectations of rewards. Government can influence the expected profitability of the inventive process through policy instruments such as preferential tax treatment for R&D expenditures, paying direct subsidies to inventors, or through the granting of property rights over invented products by means of patents. Optimal design of these policy instruments depends on an understanding of the process of invention. This project examines patent policy.

Project

The aim of this project is to test a computer simulation model of the process of invention in order to address the question:

• How does patent design affect the rate of invention (i.e. how should we design patents in order to get the fastest rate of technological progress in a society)?

In the model, a population of inventors search on a high-dimensional technology landscape for ever-better product designs. The inventors engage in personal trial and error – and in social learning from others (as captured in an evolutionary algorithm framework). Different ‘patent policies’ can be simulated – and the object is to determine what types of patent policies produce the best results.

As the simulator is already coded – there will not be a substantial ‘coding’ requirement. However, the participant should be comfortable with data analysis – and should have an interest in the idea of simulating social processes (artificial worlds!) on computer.

More generally, there is a lot of research in Artificial Life which seeks to gain understanding of real-world processes by simulating them on a computer (for example, artificial stock markets etc.). This project falls under the broad heading of Artificial Life.

Supervisors and Mentors:
Dr. Michael O'Neill, Tony Brabazon