(I spent a few days at exactly 500 wondering why I couldn't downvote yet...)
(> (karma user) downvote-threshold*)
Right you are.[0] Based on this handy Github repository: https://github.com/wting/hackernews Original: http://arclanguage.org/install
I have a couple of handles* on HN. One of them is this one, which has enough karma to downvote, and the other has too little karma to downvote. After being really thoroughly raked over the coals for a time, I began mostly logging in under my other handle in part because some people had been really horrible to me and I found myself going "YOU ARE AN ASSHOLE -- DOWNVOTE!" I didn't like interacting with the site that way and was also worried it would have negative consequences for me personally, especially since pg was not handling things in a transparent manner and there was no clear, easy way to get with a mod and sort anything out. I returned to active use of this handle at a point where I was less bitter and also the culture seemed less toxic and I no longer felt like I was just being consistently shit on, all the time. So that isn't an issue for me anymore.
And my main point is that moderating is a human thing. Karma and other technical tools are just that -- tools. There are no magic solutions. Algorithms and what not have their uses but we are all human and building community is more about how people treat each other and making judgment calls and all that then about any specific technical anything.
* This is not a secret and there is no reason to believe they cannot be readily connected to each other -- it was not done to fool anyone here, it was done for other reasons.
If they wanted to make an adjustment to the downvote karma threshold, I would suggest "require comment with down-vote" would be an interesting thing to test.
I get that not all downvotes NEED a comment categorically but there's some meat on the "downvotes deserve comments" bones.
Also: A couple times I have accidentally hit the down arrow. There is no undo. How does this relate to comment quality? The only remedy is to add a "Sorry, I downvoted you accidentally" comment, which I've seen people do. Although it's good etiquette that's a relatively low-quality comment.
I would of course assume that less people downvoting + more specific complaints would lead to more moderator action on the more obnoxious/distracting posters, if that wasn't the case I wouldn't recommend the change either.
Your karma is proof of your ability to judge a comment without the need to refute it, whereas anyone with <501 karma should only be left the opposite?
I'm aware this is how the majority of HN members use the down-vote feature, but the first time I've seen it presented as working-as-intended.
The fact that you're saying this makes me think we need more formal guidelines on the proper use of this feature. More often than not I see it claimed that down-voting should be an expression of "adds value" vs. "does not add value".
If I had to choose, I'd pick the latter.
Actually, that may not be a bad way of putting it. Comments that are blatantly false do not add value. Comments that get something wrong, but are otherwise high-quality, still add value. So this distinction may capture the nuance we want here. Let's think about it.
On a site like stackoverflow where the goal isn't discussion but answers to questions it makes sense to use a combination of mostly upvotes (for good material), lack of any votes (for mediocre material), and very occasional downvotes (for actively bad/outright incorrect answers). In that case it makes sense to engage with the poster and explain the problems, which can increase the potency of the downvote by providing context, while also helping to correct the original poster's misunderstanding.
On a site like HN that is mostly just opinion and commentary those patterns don't make sense. Most of the time comments that deserve downvoting are the sort that are either poisonous to productive conversation or just so monumentally wrong they need to go away. In that case engaging with the poster is precisely the wrong course, because doing so usually exacerbates the problem. Either by validating the behavior (if it's trolling) or by dumping a side-discussion into the larger conversation and polluting the SNR.