Comment by chimeracoder
13 years ago
> Alas I don't know for certain, but my guess is that if you have a high average karma then your posts are automatically weighted higher, so you can inject your opinion into almost any topic at any time. This affords me the luxury of being a contradictory whine even if I come late to the party!
One thing I dislike about the 'average karma' is that it discourages engaging in an extended discussion. For example, if you reply to this post, you may get a number of upvotes, but probably not as many as your original post. This creates an incentive just to ignore replies (when possible) so as not to bring down your average comment score.
It matters in high performance Python.
Just sort the comments randomly, as long as their karma is >=0, and place the negative-karma comments at the bottom.
Done. :P
This algorithmic craze is too clever for its own good. It is particularly evident in the false-positive bans.
(And I swear, it's scary how HN joke comments are indistinguishable from the real ones.)
Well, random sorting might be a bit of an annoyance due to HN's other annoyance, the "Unknown or expired link" problem, which seems to crop up more often if I do actual work and then pop over to HN when I have to wait for tests/compilation/etc. Locating the comment you were replying to with random or date-based sorting (so that you can click "reply" again, paste your previous reply, and then resubmit) would be a serious pain in combination with that.
But you're correct in that the current algorithm (or any sufficiently transparent algorithm) does lend itself to gaming. It might be harder to game if you didn't mind penalizing comments based on the quality[1] of replies they generate, then sorting based on the thread's average quality would mean that there's less incentive to post in the top thread just to get exposure, since your new, karma-less post will drag down its average and subsequently could drag down the thread. That could get prohibitively expensive to calculate for huge threads, but would be less prone to gaming, I think.
[1] If we can hand-wave "karma" to mean "quality"; I get way more upvotes when I am flippant but technically correct than I do when I try to be thoughtful.
I'm not sure how much you're kidding, but locally perturbing the comments might help a lot. Similar to simulated annealing.
1 reply →
>This algorithmic craze is too clever for its own good. It is particularly evident in the false-positive bans.
Source?
2 replies →
I have stuff to say about JIT.
Absolutely, if you care about those numbers. I think my average karma is pretty low, but that's not why I'm here.
Well, no, right now average score is meaningless. But if having a higher average score means that you have a stronger voice, as OP suggested, then people will start to care about those numbers.
I assume the goal is to encourage high-quality discussions (including extended discussions/back-and-forth), in which case average comment score will work against that.
If you measure something, people will try to increase it. Just by presenting the number, people are encouraged to maximize it.
I tend to think that better discussion in a forum like this comes about with less back and forth. That's not to say that alternating comments on opposite sides of a viewpoint can't be enlightening, but I think it's stronger if those viewpoints come from multiple people on both sides.
The metric which would be a tad closer to ideal would be some average of u/t where u := the number of users who upvoted a given comment, t := total number of users who saw that comment. Where "saw" can mean either "the comment was already there when the user loaded the comments page", or, if you want to get fancy, "the comment was within the browser's viewport long enough to be read" (which some clever JavaScript can tell you.)