Piffle. What is the "fitness cost" of a piece of code that never gets performed, or a segment of memory that never gets used? Computer simulations are not real biological worlds, where the cost of failure is loss of resource to the genome. In simulations, the cost of failure is a few immesurably small smidgens of juice from the wall socket. It isn't writ in stone that I need to punish failure with certain death, as long as I provide some differential advantage to the sucesses. It's a truism of entomology (been demonstrated in sealed mason jars thousands of times) that when two nearly identical species occupy nearly the same contained biological nitche, one or the other will eventually prevail entirely, no matter how tiny its differential advantage.
Exactly what I mean, an electronic model does not have such a fitness cost. However, in real life there is a fitness cost of non-useful organs, DNA, etc. It takes energy, food, etc. to keep such useless things alive so there is definitely a fitness cost. This is just one of the examples why this model is false. In fact, you are agreein with me:
when two nearly identical species occupy nearly the same contained biological nitche, one or the other will eventually prevail entirely, no matter how tiny its differential advantage.
Further, as I mentioned, the problem for evolution is to slowly, gradually, in small steps create a totally new organ, function, etc. with each single step making the organism more fit. This is the part of my argument you do not wish to discuss.