The PACE project has brought research on evolvable, complex systems IT into direct contact with artificial cell research and microsystem interface engineering, allowing an assessment of the potential of a combination of the latest ideas from self-organization theory with the latest technology from chemistry and MEMS. Where do we stand after this initiative? Certainly, there is much work still to be done, and the project has revealed that most of the reacting chemical systems employed are still in their early infancy with respect to technical controllability. The price of self-organization and construction flexibility is still too high in terms of loss of control and reproducibility of behavior. However, there is now a clear technological path to a whole new world of information processing - constructive self-organizing information processing - in which programmable artificial cells are a clear milestone, not far-off in the future, serving to focus technical and conceptual effort.
Von Neumann's vision of bootstrapping complexity via self-reproducing automata, while one potent factor behind the PACE project, clearly misses key features of complex systems that have proved essential to the origin of living complexity. The
vision developed in PACE also goes well beyond the ideas of Turing in bootstrapping complexity via chemical pattern formation. It is now clear from work in PACE that ongoing physical self-assembly of genetically tailored components provides a powerful basis for bootstrapping complexity and building decentralized information processing systems. A bottom-up genetic capability relies on the ability to transfer molecular information, and although this is present in very simple chemical mechanisms (template chemistry) this area needs a major technical focus to build this basic capability up into powerful information processes. The utilization of self-assembly ranges from the construction of cooperative repetitive structures (such as membranes from self-assembling amphiphiles) to very precise spatial arrangements capable of performing detailed calculations.
The metabolic turnover of components, as studied in PACE, gives self-assembly processes ongoing self-repair and adaptive properties. Resources for constructive information systems are a much more important issue than power in conventional electronic systems with fixed hardware. The research in PACE on complex resource management has revealed that it is possible with simple feedback mechanisms to build physical systems that intelligently and autonomously process specific resources, and that even a very limited range of material resources (a handful of chemical building blocks) can suffice to generate systems capable of complex self-assembly and autonomous regulation. Research on this area of fundamental primitives for constructive information processing has been integrated and accelerated by the PACE project.
PACE has revealed a programming strategy for such systems, that explicitly takes into account their self-organizing capability, their genetic base and their real-world physical embodiment. Prerequisite for an effective programming capability is the ability to interact with the chemical systems in a combinatorially complex way with feedback loops on the appropriate space and time scales. The omega machine developed in PACE demonstrates such an interaction environment for programming complex chemical systems towards the functions of artificial cells. Conventional human computer interfaces can be then linked directly to a controlled physical system that is in direct interaction with the autonomous material processing. These hybrid systems are then not just a stepping stone en route to full artificial cells, but present a first prototype of constructive computers.
While there are interesting formal issues about information processing in such systems, it has become clear that the vast majority of natural information processing occurs at the molecular scale and deals with information processes that receive direct input form dynamical structural configurations and have direct output in modulating these configurations locally. Future emerging IT will see a continuation of this trend to exploit the natural information processing capabilities of physical systems,
already a leading design principle in robotics under the name of embodiment. PACE has shown how chemical embodiment, or better "immersion", can be harnessed
in connection with the complex self-organization of chemical systems.