> Let's agree on one assumption: largely, comments
> on HN will be voted upon fairly.
I see no strong reason to make that assumption, and further doubt the existence of a universally agreed upon definition of "fairly". I think the simpler explanation is that some comments that you (and I) think are good are downvoted by others who think they are bad.Perhaps some of these are "vendetta" votes against the individual, but I think most downvotes are based on the content and tone of each individual comment. While there are sometimes extreme short-term swings to the negative, I think the process tends to balance out with extra upvotes for comments the majority considers (rightly or wrongly) unfairly grayed out.
But in my opinion, the result is majority rule, rather than anything universal or conspiratorial.
I was going to make an edit note pointing out this change since having made my comment, but chose not to.
Trust me, the rankban mechanisms exist, some really smart users did some tests to confirm it, and a lot of people are in the know by now. We had this dispute about hellbans, shadowbans, and slowbans, that was much easier to prove. Rankban has been studied pretty extensively at this point, and everyone has been convinced of it. The current effort is to study the downvoteban and policy changes.
Okay.
Let's agree on one assumption: largely, comments on HN will be voted upon fairly. That is to say, trollish comments will be downvoted (and I'm sure you know of the basics, that heavily downvoted comments will sink down). And going further in this continuum, HN voters are intelligent enough to distinguish between good-faith, substantive posts from content-free, low-substance posts -- and vote as such.
By this logic, eli's more substantive posts in a certain comment thread should be higher than its sibling-level posts. But they're consistently not. They're either at the very absolute end, or one above the absolute end.
Examples:
https://news.ycombinator.com/item?id=9430757
https://news.ycombinator.com/item?id=9421905
https://news.ycombinator.com/item?id=9402551
https://news.ycombinator.com/item?id=9370954
https://news.ycombinator.com/item?id=9371219
https://news.ycombinator.com/item?id=9358256
https://news.ycombinator.com/item?id=9346313
Do you really want me to keep going? Because this is too easy. All I'm doing is going in his comment history page, picking out comments that are substantive, good-faith, clearly non-trollish, clearly non-partisan... and I assume they should be ranked above comments that are better than his by any reasonable criteria, and consistently you find they're either not or only barely above the negatively-scored comments. You can do this exercise for any other user on rankban.