Age at Death

The graph of the age at death of programs in the population should show us whether they are evolving along the longevity axis. Figure 5.7 shows the graph for this run. There are a number of interesting points to note about this graph. First, there does appear to be some evolution towards increased longevity. On top of this, there is considerable structure in the distribution of ages at which organisms die. This is interpreted as indicating that the cycle of births and deaths in the population is highly synchronised throughout the run. The figure shows that the majority of programs live for some multiple of roughly 130 time slices at the beginning of the run, with fewer programs surviving for each successive multiple. At the end of the run, we still see the periodic structure, but the period has increased to about 160 time slices. These period lengths correspond very well with the replication period of the programs (see Figure 5.4).

A likely explanation is that the observed periodic structure arises as a consequence of the population experiencing a ceiling effect. Each time the population size reaches the ceiling, a number of programs die, creating space for the remaining programs to reproduce. Once this reproduction stage occurs, the population size is soon at the ceiling again, so the cycle repeats. The extinctions triggered by the population size hitting the ceiling are therefore periodic, resulting in the observed distribution of ages, with most organisms surviving for an integral multiple of the period of this cycle. The maximum number of programs (more accurately, cells) allowed to coexist in this run was 2500 (specified by the parameter max_cells_per_process). The actual population size never approached this limit (see Figure 5.3, discussed in the next section), so this was not the source of the ceiling effect. The other possible limiting factor is the total amount of energy distributed in the environment at each time slice. As discussed in the next section, it seems likely that this factor did indeed create a ceiling effect on population size, and therefore explains the observed distribution in Figure 5.7.