Attempts to implement molecular or nanotechnological replication systems for evolutionary engineering are confronted with the problem of exploitation by “parasites”. This is a similar problem to that confronting theories about the origin of life. Of most interest to evolution technology are informational molecules R catalyzing replication processes in the form R + X → R + X + X (co-operative replication), using X as template to produce a copy of X, where X represents a replicable entity out of a combinatorially large class.
In an ideal replication system, there is
only one type of X, namely R. However, in order to be of interest for evolving
functional X, the replication ability of R must be of some generality,
otherwise the best one can hope to implement is an (auto-catalytic) replication
system without any possibility of further development, and even this proves
impossible for complicated R, since accurate recognition of all pertinent
information can never be complete for large R. In other words: The ability to
evolve requires a generous replicator, not complete template recognition.
Physical replication is generically error
prone: Even if one starts with a population consisting solely of R’s, via
mutation, individuals arise that have lost any functionality but may well be replicated
further. They consume system resources without contributing to the overall
replication. Some of these “parasites” will have a higher replication rate, e.g
because they are smaller.
Using such arguments it is straightforward to show that replication systems as we know them cannot be maintained in a well-mixed population where generic parasites can occur. Spatial structure and fluctuations owing to small local population sizes have been recognized as physical mechanisms that can stabilize a self-encoding catalytic replication system against exploitation.
Such spatial combinatorial kinetic processes are otherwise difficult to simulate over long evolutionary time scales, simply because of their computational demands. Furthermore, single Monte Carlo simulations do not give analytical insight into the interplay of different system parameters. PRESS simulations and theory solve this problem.