Clojurians Log v2

RE my book recommendation from last week's meetup: https://www.amazon.com/Why-Greatness-Cannot-Planned-Objective/dp/3319155237 I just finished this. It's good, though it gets repetitive, the major points can be gleaned from the first few chapters. They're point in brief: searching for novelty is often a better approach to solving a problem then explicitly working toward an objective. They compare novelty search with standard stochastic gradient descent for solving AI problems, and find novelty searching beating SGD in many cases. After finishing, I thought maybe the author's had overstated their case. Then I saw that one of the author's has continued investigating the "Novelty Search" machine learning algorithm, and discovered that the layers of the neural networks that their algorithm created had a more human understandable structure (what they call unified factored representation). Their paper is here: https://github.com/akarshkumar0101/fer?tab=readme-ov-file Very interesting stuff.