Mutations and Flaws

Little has so far been said about the role of mutation in Cosmos. Mutation is a vital process from the evolutionary point of view, as it provides a continual source of genetic novelty for selection to work upon. Mutations occur naturally throughout the system at a low rate, and may affect most of the structures within the cell (i.e. the Genome, the Received Message Store, the Nucleus Working Memory, the Communications Working Memory, the Promoter Store, the Repressor Store, the flaw rate, the stack, the registers and the flag). For structures which are based upon BitStrings, mutations are governed by the global parameter mutation_period, which specifies the probability of an individual bit within the structure being flipped. For structures based upon integer numbers (the flaw rate, stack and registers), mutations occur at the same rate as for BitStrings, but the details are slightly different. For the flaw rate, a mutation causes a random increment or decrement in the current value within predefined limits.^4.17 For the stack, a mutation will, with equal likelihood, either cause a random number to be pushed onto the stack, or the top number to be popped off it. For registers, a mutation will cause the register's current value to be replaced by a random value. Mutations also affect the cell's flag at the same rate, causing the flag's state to be inverted.

In addition, variety may also be introduced into an organism by the flawed execution of instructions in its genome.^4.18 When a flaw occurs (which happens at a rate defined by an individual cell's flaw rate, as described in Section 4.3.8), the instruction which is about to be executed, rather than just being executed once, will either be executed twice (successively) or not at all. (The choice is random, with both events occurring with equal likelihood.) The effect of a flaw is therefore that instructions may occasionally produce abnormal results, such as an inc_a instruction adding 2 to the value of the ax register instead of 1.

Despite this distinction between mutations and flaws, the net results are the same. If the error affects what gets written to the Nucleus Working Memory of a cell just before it issues a nwm_divide instruction, then it will be passed on to the child organism and become a permanent addition to the gene pool. On the other hand, if the error does not affect the contents of the Nucleus Working Memory (even indirectly), and it does not affect the regulators that get passed on to any offspring, then it will only affect the current organism and will not be inherited by child organisms. From an evolutionary point of view, only the former scenario is important.