Rarity and mining for gems in the edge cases

Excerpt below from my Medium essay NFT features: An opportunity to provide a rich and expressive title substitute for long form generative artworks

When first approaching the task of assigning features for this project I originally thought that because the rarest attributes were probably going to be valued the most, that the real magic should happen in the probability spaces defined by those attributes. Any generative artist will tell you however that the real magic is not something that’s easy or possible to predict. When we can know how to create it then we evolve our algorithm in that direction and then from this we will get a new subset with a new level of real magic. The understanding of how to capture the very highest expressions of our algorithms is always just out of reach. The flip in perspective I had was to say that the rarest attributes should correlate with the edge cases where the most wild outputs live. Amongst those wild outputs there can be both the gems and the duds. The middle of the bell curve of results I want to be filled with my most consistently reliable outputs that I am very happy to represent my project as a whole. Many of the visual trades I like the best are reserved for this middle ground as this is the meat and veg of my project and you want to give your best to the most buyers. To have the possibility of unearthing some iconic gems in new unstable seams, there must also be the possibility of a mining disaster. To use a market analogy, the rarest attributes in this approach define the space where the rewards are subject to the greatest volatility.