> Could code complexity be measured as cumulative entropy as measured by running LLM token predictions on a codebase? Notably, verbose boilerplate would be pret

> Could code complexity be measured as cumulative entropy as measured by running LLM token predictions on a codebase? Notably, verbose boilerplate would be pretty low entropy, and straightforward code should be decently low as well. WinRAR can do that for you quite effectively.