This article argues that any conceivable society that holds art competitions will be less capable than a society that does not, because art does not represent real world objects so holding a competition in art trains people to lose track of reality, eventually rendering them unable to do even basic functions as they try to paint instead of walk.
After all, if you aren’t currently looking at a flesh and blood rhino, you can never learn to get better at drawing rhinos, and that shoddy workmanship means you’ll never be able to draw a savannah (which has rhinos and many other things), let alone develop the live-updated 3D model you need to walk across one.
Soon, people who aren’t currently looking at rhinos will be unable to drag themselves across the floor, not knowing how to draw the hands and feet that would propel them.
The article makes general assertions about systems. It gives plumbing, software, and machine learning as examples, and I simply add the examples of human learning and human societies.
An agent specifying the criteria of a test inspired by the real world and then evaluating the performance of a bunch of other agents is equally information-incestuous when those agents are LLMs as when they are humans. If the observations of the article must hold true for LLM training, then they must hold true for humans. If they do not hold true for humans, then they do not prohibit the existence of LLM training programs that likewise avoid the decoherence trap.
Given the fact that evolution was able to find human creativity and art as forms of cohering enrichment of human performance beyond the limits of the training data of a world without art, I would expect that there is a good chance tens of thousands of engineers and billions of dollars of compute will be able to find some form of abstract expression and learning as a form of cohering enrichment of LLM performance beyond the limits of the training data of a world without that abstract LLM expression.