H1 That conventional methods can only work within constrained regions of electronics design space has been shown first by characterising what conventional design practices actually are, and secondly by exhibiting an evolved circuit obviously beyond them: the tone discriminating circuit of Case Study 2.
H2 That evolutionary algorithms can explore some of the regions of design space that are beyond the scope of conventional methods has been shown explicitly. In Case Study 1, the DSM robot controller, evolution was seen to exploit the enhanced behavioural capabilities of a conventional architecture, once some constraints from digital design methods were relaxed. In Case Study 2, given the freedom, an evolutionary process produced a working circuit with a structure and dynamics foreign to orthodox design and analysis.
H3 That evolutionary algorithms can in practice produce designs beyond the scope of conventional methods, and that are better, has been signalled repeatedly but not demonstrated conclusively. On theoretical grounds, there is greater opportunity to find the forms and processes that naturally exploit the properties of the electronic medium, and this seems to happen in practice. Nonbehavioural requirements can be integrated into the design process more seamlessly, and an example was seen when the robot controller was evolved to be tolerant to SSA faults, without the explicit incorporation of redundant parts. There is the promise of the evolution of means for robustness that are tailored to the task, the circuit, and the medium. Hardware and software tools were presented to illuminate the way forward in scientific investigation. H3 is at least a good working hypothesis.
What initially seemed daring hypotheses are now either matter-of-fact, or within reach. From the vastness of design space, practically useful novel regions beckon.