Post

I am proposing the 10:1 rule for LLM generated code. For every token you used on the writing, you should expend 10 times as many on the review.