4 ms·

I think this change makes the mistake that people with lots of karma are good contributors. I think it should be based on weighted karma/comment in addition to

I think this change makes the mistake that people with lots of karma are good contributors. I think it should be based on weighted karma/comment in addition to total karma. Imagine two users - one with 1500 karma & 1.1 karma/comment versus one with 600 karma & 10 karma/comment, which one would you trust to be able to judge what is a good comment and what isn't? The total karma weight is to give at least some favor to frequent users, but not too much. In addition, if someone replies to your comment, you should be able to at least see their comment regardless of your karma or whether it's been greenlit.

tedsanders 13y ago
That's a good point, but I think it would have an unintended negative consequence of punishing late comments, which tend to have low karma but may nonetheless be worthwhile. It would also punish comments on 'new' threads, which often stagnate with 0 or 1 upvotes.
dudus 13y ago
Karma/Comment is not a very good metric as well. As someone that posted a big post that stayed in the Front Page for long can amass a lot of karma in little time. Giving them a very high Karma/Comment. You would have to remove the outliers.
- jamesaguilar 13y ago
  A perfect opportunity to use comment-karma H index! http://en.wikipedia.org/wiki/H-index http://en.wikipedia.org/wiki/H-index
- corresation 13y ago
  Karma/Comment is a poor metric, I think, because it encourages the strike attack comment style. If I post something and people post counterpoints or questions or concerns, I will respond. But the truth is that from a pure karma average perspective, by far the best action would be to simply leave the threads hanging, the single post floating at the top of karma heaven. That works great from a pure karma perspective, but it subdues conversation because every point is a single attack. It also encourages the stealthy comment editor. While Paul mentions this, it is something that happens regularly -- people make a comment, someone responds unfavourably, and the original person subtly edits their comment to make them look unfairly attacked. This is a specific problem on HN given that quoting is generally discouraged. So everyone piles on the downvotes and upvotes, respectively, to right this seeming injustice. Karma on HN is easily gamed, as it is elsewhere. It is unfortunate when we care too much about it because it leads to completely artificial conversations.
tokenadult 13y ago
I think this change makes the mistake that people with lots of karma are good contributors. That's not an invariant property of participants with karma over 1000 points, to be sure, but the population of participants with over 1000 points skews to include a fair number of people who have been long-term contributors (patio11, who has my same join date and deservedly much more karma, immediately comes to mind among several other participants). What's practical about this bright-line rule is that it strictly limits the number of participants who can review pending comments at all, and allows the moderation team here to go to the next step of evaluating the reviewers, while ALL OF US are put on notice that what's desired here is substantive, polite comments. So while I agree with you that the relationship "high karma implies good contributor" is not an invariantly true relationship, it has enough heuristic value to start as a seed for the system as further hand-tweaking of review power by the moderation team responds to your legitimate concern here.
ilyagr 13y ago
I think that bsamuels' point about replies is also important: I sometimes reply to comments with the intention to be helpful to the author of the comment, not necessarily everybody else.
thaumasiotes 13y ago
> I think this change makes the mistake that people with lots of karma are good contributors. You're making the related mistake that comments with lots of karma are good comments. By coincidence, I (at the time of writing) have roughly 1500 karma at 2.2 / comment. Here's a selection of some of my sub-10-karma comments that I happen to like: https://news.ycombinator.com/item?id=7203273 https://news.ycombinator.com/item?id=7203273 (for reasons I don't understand, 0 karma) https://news.ycombinator.com/item?id=7276538 https://news.ycombinator.com/item?id=7276538 (2) https://news.ycombinator.com/item?id=7156141 https://news.ycombinator.com/item?id=7156141 (3) https://news.ycombinator.com/item?id=7396659 https://news.ycombinator.com/item?id=7396659 (4) https://news.ycombinator.com/item?id=7239733 https://news.ycombinator.com/item?id=7239733 (4) https://news.ycombinator.com/item?id=7352327 https://news.ycombinator.com/item?id=7352327 (5) Here's every 10+ karma comment I've made in the past 50 days: https://news.ycombinator.com/item?id=7206460 https://news.ycombinator.com/item?id=7206460 (10, with a 12-point comment two levels down) https://news.ycombinator.com/item?id=7329150 https://news.ycombinator.com/item?id=7329150 (12) https://news.ycombinator.com/item?id=7296981 https://news.ycombinator.com/item?id=7296981 (11) https://news.ycombinator.com/item?id=7206436 https://news.ycombinator.com/item?id=7206436 (14) https://news.ycombinator.com/item?id=7194772 https://news.ycombinator.com/item?id=7194772 (22) Weirdly, the intuition I got from going back over these is that I'd probably use a threshold of 8 to divide "popular" comments from "unpopular" ones. Anyway, I see a clear trend in the 10+ comments: they either openly mock someone, or they take a stand on a contentious political issue (the 22-pointer does both!). My favorite of them all is the 5-pointer, and I dearly hoped that someone would reply to it... but no. Obviously, I've got plenty of less-worthwhile 1- and 2-point comments to my name. But I'm not at all convinced that someone with 10 average karma per comment is making better comments than someone with an average of two; it seems more likely that they're either demagoguing or just witty, possibly nastily witty. postscript: average karma of some bright lights: patio11: 15.2 cperciva: 11.3 tokenadult: 6.1 tptacek: 5.4

tedsanders 13y ago

That's a good point, but I think it would have an unintended negative consequence of punishing late comments, which tend to have low karma but may nonetheless be worthwhile. It would also punish comments on 'new' threads, which often stagnate with 0 or 1 upvotes.

dudus 13y ago

Karma/Comment is not a very good metric as well. As someone that posted a big post that stayed in the Front Page for long can amass a lot of karma in little time. Giving them a very high Karma/Comment. You would have to remove the outliers.

jamesaguilar 13y ago
A perfect opportunity to use comment-karma H index! http://en.wikipedia.org/wiki/H-index http://en.wikipedia.org/wiki/H-index
corresation 13y ago
Karma/Comment is a poor metric, I think, because it encourages the strike attack comment style. If I post something and people post counterpoints or questions or concerns, I will respond. But the truth is that from a pure karma average perspective, by far the best action would be to simply leave the threads hanging, the single post floating at the top of karma heaven. That works great from a pure karma perspective, but it subdues conversation because every point is a single attack. It also encourages the stealthy comment editor. While Paul mentions this, it is something that happens regularly -- people make a comment, someone responds unfavourably, and the original person subtly edits their comment to make them look unfairly attacked. This is a specific problem on HN given that quoting is generally discouraged. So everyone piles on the downvotes and upvotes, respectively, to right this seeming injustice. Karma on HN is easily gamed, as it is elsewhere. It is unfortunate when we care too much about it because it leads to completely artificial conversations.

tokenadult 13y ago

I think this change makes the mistake that people with lots of karma are good contributors. That's not an invariant property of participants with karma over 1000 points, to be sure, but the population of participants with over 1000 points skews to include a fair number of people who have been long-term contributors (patio11, who has my same join date and deservedly much more karma, immediately comes to mind among several other participants). What's practical about this bright-line rule is that it strictly limits the number of participants who can review pending comments at all, and allows the moderation team here to go to the next step of evaluating the reviewers, while ALL OF US are put on notice that what's desired here is substantive, polite comments. So while I agree with you that the relationship "high karma implies good contributor" is not an invariantly true relationship, it has enough heuristic value to start as a seed for the system as further hand-tweaking of review power by the moderation team responds to your legitimate concern here.

ilyagr 13y ago

I think that bsamuels' point about replies is also important: I sometimes reply to comments with the intention to be helpful to the author of the comment, not necessarily everybody else.

thaumasiotes 13y ago

> I think this change makes the mistake that people with lots of karma are good contributors. You're making the related mistake that comments with lots of karma are good comments. By coincidence, I (at the time of writing) have roughly 1500 karma at 2.2 / comment. Here's a selection of some of my sub-10-karma comments that I happen to like: https://news.ycombinator.com/item?id=7203273 https://news.ycombinator.com/item?id=7203273 (for reasons I don't understand, 0 karma) https://news.ycombinator.com/item?id=7276538 https://news.ycombinator.com/item?id=7276538 (2) https://news.ycombinator.com/item?id=7156141 https://news.ycombinator.com/item?id=7156141 (3) https://news.ycombinator.com/item?id=7396659 https://news.ycombinator.com/item?id=7396659 (4) https://news.ycombinator.com/item?id=7239733 https://news.ycombinator.com/item?id=7239733 (4) https://news.ycombinator.com/item?id=7352327 https://news.ycombinator.com/item?id=7352327 (5) Here's every 10+ karma comment I've made in the past 50 days: https://news.ycombinator.com/item?id=7206460 https://news.ycombinator.com/item?id=7206460 (10, with a 12-point comment two levels down) https://news.ycombinator.com/item?id=7329150 https://news.ycombinator.com/item?id=7329150 (12) https://news.ycombinator.com/item?id=7296981 https://news.ycombinator.com/item?id=7296981 (11) https://news.ycombinator.com/item?id=7206436 https://news.ycombinator.com/item?id=7206436 (14) https://news.ycombinator.com/item?id=7194772 https://news.ycombinator.com/item?id=7194772 (22) Weirdly, the intuition I got from going back over these is that I'd probably use a threshold of 8 to divide "popular" comments from "unpopular" ones. Anyway, I see a clear trend in the 10+ comments: they either openly mock someone, or they take a stand on a contentious political issue (the 22-pointer does both!). My favorite of them all is the 5-pointer, and I dearly hoped that someone would reply to it... but no. Obviously, I've got plenty of less-worthwhile 1- and 2-point comments to my name. But I'm not at all convinced that someone with 10 average karma per comment is making better comments than someone with an average of two; it seems more likely that they're either demagoguing or just witty, possibly nastily witty. postscript: average karma of some bright lights: patio11: 15.2 cperciva: 11.3 tokenadult: 6.1 tptacek: 5.4