The project described here has two milestones of success: 1) Survival of the differentiated state under conditions of free evolution by natural selection. 2) Increase in the number of cell types through evolution. Before the bugs in the sensory system were fixed, the results were negative for both milestones. Once the sensory system became fully functional, the first milestone, but not the second, was achieved.
The dysfunctional sensory system is rapidly eliminated by natural selection, usually within the first twenty-four hours of evolution, suggesting that it does not enhance the fitness of the organism. The functional sensory system however, persists indefinitely, suggesting that it enhances fitness.
In the absence of a fitness contribution from the sensory tissue, it turns out that an undifferentiated algorithm with a random dispersal strategy is quite effective, and survives preferentially over the more complex algorithms that ineffectively attempt to use differentiated CPU threads to identify the nodes with a higher ratio of Speed/NumCells.
The success of the differentiated form with the functional sensory system suggests strongly that the ability to obtain and process information about conditions in the environment, and to use that information to direct movements within the environment enhances fitness. The fitness benefit of the sensory system evidently is great enough to counterbalance the cost of maintaining the additional genetic information, developmental process and tissue.
This result forcefully states the challenge of this work: To find the conditions under which the more complex differentiated organism can survive under conditions of free evolution by natural selection, in potential competition with the simpler undifferentiated organism.
Finally the current work is explicitly not about the conditions under which the differentiated form might arise spontaneously. Rather it is about the conditions under which an already primitively differentiated form (with just two cell types) can survive, and perhaps increase its level of differentiation through evolution. This is analogous to the original Tierra experiment which was seeded with already functional replicators. That experiment was explicitly not about the spontaneous origin of replication and evolution, but about the evolutionary changes that occur in already existing replicators. Subsequently, Pargellis [4, 5] was able to study the spontaneous origin of replication and evolution in a derivative of the Tierra system. In general, it is probably best to establish that a property can be supported in a system, before attempting to study its spontaneous emergence.