I suggest that the most likely way to achieve complexity increase in digital evolution is through evolution by natural selection in an ecological community. No attempt should be made to provide fitness functions, or artificial selection, to guide evolution towards useful products. Rather, evolution should be free to explore the possibilities without the burden of human ``guidance''.
Traditionally we have managed evolution through manipulating selective forces. In this new approach, our role is to create the conditions for complexity increase, rather than trying to guide it through artificial selection. This is an interesting scientific challenge, as the conditions that generate complexity increase are unknown. However, the process of attempting to generate a complexity increase amounts to an experimental approach to the problem. We can try many different approaches, and if we are successful, we will be able to experiment with the system until we can find the minimal conditions that generate the behavior.
Above I have discussed some ideas on how the environment can be manipulated to encourage complexity increase (the biodiversity reserve for digital organisms on the internet). But there are many other aspects of the design of the system that can be manipulated. One of the most important of these is the design of the genetic language on which evolution is based. The language issue was discussed in the section ``Evolvability and the Language'' above.
In addition to the design of the environment, or the genetic system itself, there are other features of the system we can manipulate. For example, we can also design mechanisms to facilitate gene regulation in multi-cellular digital organisms. Or we can design mechanisms to facilitate an organized sexuality comparable to the sex in organic life that underlies the species/gene-pool phenomenon. It is the design and introduction of these kinds of elements that constitute the role that we can play in attempting to ``manage'' the evolution of complexity, as opposed to our traditional role of only providing the selective forces.
This process of attempting to engineer an evolutionary increase of complexity, amounts to a experimental approach to the study of the evolution. If the efforts are successful, the empirical results from these experiments can support the development of new theoretical underpinnings for evolutionary biology.