Over the last few years, I’ve been working on Zyme, an esoteric language for genetic programming: creating computer programs by means of natural selection. I’ve started seeing promising results, showing that random bytecode mutations can, over time, lead to measurable improvements in program performance. While still a long way from state-of-the-art approaches like neural networks, I wanted to share my progress in a blog post.
Feedback and criticism are welcome!
