News for hackers. Stuff that matters.
Re:Ask HN: A Karma API? by rms on Saturday November 29, 2008 @02:43AM (#379977)
Let's start by allowing karma transfers between users here...
Re:Ask HN: A Karma API? by nebula on Saturday November 29, 2008 @10:22AM (#380145)
While going through some of the recent controversial threads on HN, I felt that something like this would be a nice thing:

When people mindlessly downmod some controversial but valid comment, some people who stand for free speech and connect with the validity of the comment might want to upmod it more than +1 and show their support. In such cases, the extra upmods should be in the form of karma transfer. It's analogous of throwing your reputation on the line and standing up for someone.

Re:Ask HN: A Karma API? by gojomo on Saturday November 29, 2008 @03:39AM (#379998)
Or how about transfers just between our own comments?

I'd love to be able to slide surplus karma from dumb-but-popular things I've said to the underappreciated, more controversial comments I'm more proud of. :)

Re:Ask HN: A Karma API? by nebula on Saturday November 29, 2008 @10:12AM (#380141)
With a transfer cost of course :) Otherwise you could misuse your karma to move your comments up all the while.
Re:Ask HN: A Karma API? by Shamiq on Saturday November 29, 2008 @12:17PM (#380211)
And I have a hunch that high ranking comments gain points exponentially, not linearly.

I'm not sure if that's what I mean -- but some posts seem to get more points just because they already have a lot of points.

Re:Ask HN: A Karma API? by kragen on Monday December 01, 2008 @12:35AM (#381420)
Like pop songs in that study where they divided people into groups and had them rate songs? It seems almost inevitable. That sort of positive feedback can make a system more sensitive, but it can also make it unstable, amplifying noise until it overwhelms small signals. My third connection is that, if your hypothesis is correct (and I'm pretty sure it is) you would tend to expect comment karma to be distributed according to a Zipf distribution.