MarkvW said:
Mutation makes it inevitable that the 'whole system' will change. Change is the only certainty. We can arbitrarily define a system by genus, species, etc., but every member of that system, and all its descendants will always be mutating and passing on some of those mutations.
You can say that a functioning system is extremely unlikely to change only if you restrict membership in that system to organisms that meet your systemic definition. Those systems won't change only because you have defined-out noncompliant descendants.
My post above covers most of this, but change is incremental - one improvement happens at a time. We really don't mutate all that much, it is only over the super-long-term that changes can happen. The eye is an incredibly complex bit of biochemistry developed over billions of years, but what was an advantage a billion years ago can be something which could have a theoretically better-functioning replacement, yet the chance of you happening upon several (maybe even hundreds) of simultaneous mutations required to make a change that would actually make the new one better than the old, rather than being a kind of halfway point that did neither, is minute.
You then have odd situations like that of sickle-cell disease. If you have one "sickle-cell" allele, you are immune to malaria. Two, and your red blood cells are crescent-shaped, which is terrible for your health. The allele originates in tropical Africa, where an immunity to malaria is hugely beneficial, so the allele is propagated, but this has the knock-on effect of a large amount of sickle-cell disease. It could be possible to have the reverse situation - two would be good and one would cause death before mating age. It'd be incredibly unlikely to have both alleles mutate to the same things at the same time, and even then, you'd need a mate who had the exact same thing happen.
"if you restrict membership in that system to organisms that meet your systemic definition"
Can you clarify what this means? I can't make it out, sorry.
My overall point: evolution, via incremental change, tends to produce good (or even very good) solution, but it's highly unlikely to always produce the best result.