Smaller the function, less ways to screw it up, for humans and llms alike. I suspect humans don't code like this because it is too tedious but LLMs don't have such problems