Its not the exact opposite of what ive been reading. Basically every person claiming to have success with LLM coding that ive read have said that too long of a prompt leads to too much context which leads to the LLM diverging from working on the problem as desired.