I like to say that "stability is its own reward". The idea behind this is that many unstable plastic systems continuously adapt until they stabilize. Once stable, plasticity is limited because it was the instability that was driving them to change.
Usually my context for this is in the realm of neural network learning algorithms. Some years ago, however, I identified a couple of instances of self-stabilizing systems that exist in the real world. Each of these tongue-in-cheek examples might also be slightly cyclical -- and cynical.
First
example: clearing traffic accidents increases traffic speed. As lanes are re-opened, average traffic speeds can jump from a rush hour stop-and-go crawl up to maximum posted speed limits. As speeds increase, however, accidents also increase -- which force commuters back down to speeds that are safe for congested driving conditions.
Second example: state lottery revenues fund education. As participation in the lottery increases, education increases. As education increases, participation in the lottery decreases -- because educated people do not buy lottery tickets.