This micro-controller, one out of
a whole hierarchy of controllers with differing complexity, shows the features
of this approach. Each instruction is subjected to a condition (in the current
case, 8 conditions can be expressed). The zero-flag of the accumulator is the
most important one. Via the flags PF0, PF1 and PF2, external signals can be
detected.
The case 'never' is reserved for special functionalities such as
§ presetting the accumulator with a certain value or
§ loading a random value to the accumulator.
Here, eight special functions can be realized. 'Store' and 'Load' and the Boolean operations are defined as usual. The 'Shift'-instruction realizes a cyclical one-bit-shift. The 'Connect'-instruction is special compared to standard micro-controllers. When executed, it tries to catch a neighboring micro-controller which exhibits the recognition-site given in the Identifier-section and is masked via the wild-card-register. The Identifier-section provides the parameter for each instruction.
The evolutionary analysis of this micro-controller (placed in a certain environment) is time-consuming and difficult. The major problem, aside from the computational demands, is the analysis of the evolving programs. It is not known, a priori, what successful distributed programs look like. It is also extremely difficult to debug such a program because of non-deterministic program (control and data) flow.
Currently, pattern recognition is the preferred analysis method, because if something replicates successfully it should show up as many repeats in the genome. It is very probable (and indeed the case in real biological evolution) that conserved regions hint to important information. How to analyze these repeat structures is part of our ongoing research.