So I doubt if reading the top comments is a very objective method for evaluating controversial discussion (it has a strong correlation, but maybe not the best). Often, I see very heated discussion and competing comments moving up and down until nobody is interested in spending more energy in the debate.
P.S: Invasive profiling and tracking can be a very effective (and possibly, the only) method to uncover insights on the dynamics of online forums like Hacker News. If we track users' every move, it could make great contribution to sociology and psychology researches, and may even help answering unsolved questions in order to building a better community for everyone. Unfortunately, it's too dangerous and unethical to use, I won't support it, but I'm always curious to know the results.
Karma doesn't affect ranking on HN. This comes up often enough that I wonder where the idea came from. Do other forums work this way?
Interestingly, I originally believed HN is pure-upvote based, then I learned it from other HN comments that says karma affects ranking and I believed it.
So I'd say it's just an unsubstantiated rumor/misinformation getting circulated in the comment section from time to time, combined with the impression of HN having a "magic algorithm", so many believed it without any fact-checking. Also, the quick-moving nature of HN comments somehow created a confirmation bias that makes the idea appeared to be true.