I like your point that it's necessary to figure out the math of replication first and then apply the knowledge in design of alternative computer-based replicato

I like your point that it's necessary to figure out the math of replication first and then apply the knowledge in design of alternative computer-based replicators. In addition to the approaches from top and bottom there is one more (bottom-bottom) approach: turn an existing cellular automation like Conway's Game of Life into an evolving computer-based replicator. As far as I know, all existing examples of cellular automation can only produce replicating systems. They are not evolving because they don't have selection. The moment one of these systems is tweaked so that it would also have a possibility to mutate and endure natural (artificial would be a better word here) selection, there will be evolving life in a computer.