Evolution will treat hostility with hostility

Here's an idea that might apply to all three of humans vs. nature, humans vs. humans, and disease vs. humans: A system in which one group negatively affects the survival of another group is not stable, even if the hostile group attempts to keep it stable. A hostile entity must either completely eradicate another, or the other will evolve to disrupt the system. This would predict that humans cannot indefinitely harm nature without nature putting a stop to it, and that murderous tyrants cannot maintain power over oppressed people, and that since we try to kill all germs, superbugs will evolve to kill us.

Why? First let's assume that group A has a negative influence on the survival of group B, but with an intention to control its survival rather than wipe it out. Then, any evolved behavior in group B that circumvents death by group A, is an evolutionary advantage. So on the surface, behavior that allows B to "get along" with A is an advantage, but unless it is effective enough to disrupt the system (and make A no longer a negative influence on B's survival) thus making it an unstable system, then it is not good enough to prevent A's influence. It's not good enough for B to change its behavior to adapt to A, because A can also adapt, so if its hereditary advantage is to oppress or control B, A may also evolve to maintain control. This is what should happen if the system is evolving and stable. For example, if new superbugs evolve ways to survive disinfection, we will look for new ways to kill them. So unless the system becomes symbiotic, it is an insufficient evolutionary advantage for B to only find a way to put up with A. A better advantage would be to disrupt A, and disruptive evolved behaviors may provide the only way for B to ensure its survival. It either dies by group A, or it stops group A.

This means that superbugs aren't busy evolving a way to avoid being killed by us, they must be evolving a way to kill us, because only the group that does so will survive. A disease that can take us down will be more successful than a disease that can survive as we look for new ways to kill it.

As per the other examples, it would mean that humans cannot be sustainably harmful to nature, without either destroying it completely or inducing evolution that is harmful to humans. It also suggests that murderers are never really safe. In a stable system, neither group must be trying to kill the other, because only then would there be no certain evolutionary advantage to killing the other first before they kill you.

The hypothesis assumes that such system-disrupting evolved behaviors are always possible, and likely enough to rely on one happening eventually, but I think it's true of the examples given at least.