Towards a Theory of Open-Ended Evolution

This paper is an effort to begin serious work toward a theory of open-ended evolution (OEE). We begin by defining OEE to mean continued innovation. With this intuitive definition in mind, we closely examine metrics for characterizing behavior of evolutionary systems and argue that none are sufficient for quantifying open-endedness, but there are still avenues to explore. Working from our relatively simple definition, we are nonetheless able to reduce our root requirement of continued innovation into three, slightly more specific requirements, summarized as: 1) potential for innovation 2) support for innovation and 3) “reachability” of innovation. We then discuss how these requirements might be broken down further. In particular, we take up the question of whether the biosphere faces qualitatively different challenges in meeting these requirements than do systems like technological evolution. Our analysis seems to indicate that these seemingly dissimilar systems face identical challenges in meeting our OEE requirements (albeit on different scales), and we use this idea to aid us in attacking the third requirement—reachability—which seems to be the most challenging to ensure. We argue that two attributes: coevolutionary parts-sharing, and a “helpfully constraining” development process, may be important for establishing reachability.