next up previous contents
Next: Summary and Discussion Up: Inoculation with Sexual Ancestors Previous: Inoculation with Sexual Ancestors

Results

Analysis of the output data revealed one basic result; the capacity for sexual reproduction was lost at a very early stage in all nine runs. By studying some of the individual genotypes from the runs, it was apparent that a minor mutation in the male section of the code responsible for collecting female `spores' from the environment (the section labelled seek_female_msg in Figure 6.30) was sufficient to turn the male into a fully functional, independent, asexual organism. Mutations of this sort led to the sexual organisms being completely displaced by asexual organisms within the first 90,000 time slices in all runs, and indeed much earlier in most of them.

After the transition to asexuality had been made, subsequent improvements in the organisms were possible by the loss of the now redundant female section of the genome, and also by the loss of redundant sections of the male section (mostly within the areas labelled wander and seek_female_msg in Figure 6.30). The state of the system at the end of each of the runs can be classified as follows: in Runs 5 and 8, no progress had been made in removing redundant code from the asexual programs--they were still of the same length as the original sexual ancestor; in Runs 1, 2, 6, 7 and 9, most of the female section of the code had been removed from the programs, leading to programs with lengths in the range of 936-1098 bits (although in Run 7 this improvement only occurred within the final 50,000 time slices); and in Runs 3 and 4, parts of the redundant areas of the male section had been removed in addition to most of the female section, leading to programs with lengths in the region of 580-670 bits.

A final point to note is that in the programs alive at the end of Run 4, there was also evidence that a cancer had invaded the population. This was of the type where the extra instructions accumulating at the ends of the programs were energy collection instructions, as observed in a previous run discussed in Section 6.5.2. It is interesting that in both of the experiments in which this phenomena was observed, the parameter energy_collection_scheme was set to private. Recall that this means that programs can only extract energy from the (abiotic) environment, and not from their neighbouring programs. If such a cancer were to arise in a run in which programs could extract energy from their neighbours, the system would soon fall into Class 1 dynamics, as the amount of energy extracted would soon rise to a level where all such neighbours were killed. It is only when programs are prevented from extracting energy from their neighbours that this cancer can develop to the extent that it did in Run 4 of the current set of runs, and in Run 9 of the Private Energy Collection runs reported in Section 6.5.2.


next up previous contents
Next: Summary and Discussion Up: Inoculation with Sexual Ancestors Previous: Inoculation with Sexual Ancestors
Tim Taylor
1999-05-29