I agree. I was actually making a remark with this respect in my post.

Using comments may be the key, but they should not be used as such to. Some comments are there only to thank the author for sharing (or similar) while others are really triggering discussions. For the moment, I have no automatic way to assess what is what, beyond using the length of the comment (but is this reliable?).

What also reinforce what you said is that the distribution in the number of votes seems similar whatever is the category. I have not tried, but I guess we can assume that the distribution may be similar in mot categories (above a certain popularity threshold).