Member-only story
Rabbit Plagues, Genetic Algorithms
life finds a way
What if I told you we have a massively complex, multi-threaded optimization function simultaneously running on trillions of distributed biological computers that update their own hardware and software — and it’s all solar powered. Sounds too good to be true.
It is. That function’s called evolution, and it doesn’t seem too compatible with our computers (without human intervention designing better computers).
However, we have some ways of emulating the efficiency of evolution: Genetic Algorithms are a way to emulate natural selection using math.
The evolution I’m fond of is punctuated equilibrium, which is much more rational than gradualism. Species don’t change much unless they need to, because there’s no room or reward for doing so. An environment-transforming event occurs, and in the ensuing vacuum, change happens rapidly until efficiency plateaus are reached.
Now consider a maximization/optimization function that you ran once, and it gives you its best solutions. Run it again, and it won’t change (probably). Change the parameters, and suddenly there’s room for change.
