Hacker News Karma Tracker

The problem with karma implementations is that they don't always encode for the best outcome. I am not sure what karma means on HN. There are some topics where you could very easily fish for karma by going with the prevailing thoughts in any given community. Post against the grain and loose karma. It's fairly predictable in most communities.

Then there are effects as I've seen in some of the Stack Exchange sites. One that comes to mind is ServerFault. Moderators and high-karma members with elevated privileges have turned that particular site into a nearly unusable and hostile environment for anyone outside a very narrow definition of what the "gang" wants.

This is an interesting effect because it encodes what a few privileged users want as opposed to what --conjecture-- a larger majority might be looking for out of the site. And, from a strict business view of the problem it reduces that audience and the applicability of the site/s by a potentially significant factor.

It's like having a mechanical engineering site where high-karma users have decided they only want to discuss hex bolts used by aircraft mechanics. You show-up to ask about hex bolts used on motorcycles and watch out!

Perhaps the ultimate questions --on a site-per-site basis-- might be:

  - What's the design intent behind karma or user reputation?  
  - What is it you want to encode?
  - How do you ensure that this is, in fact, encoded into the rating?
  - How do you prevent users from polluting or de-valuing karma through
    misinterpretation, petty/spiteful voting, "gang-like" behavior or other
    degenerations?
  - If karma is used to elevate user rights, how does it also affect
    responsibilities?  Does it impose a level of accountability?
  - How do you keep karma "wars" from affecting your community?  By this I 
    mean gang-like behavior by regulars who'll attack outside their pack
    with the only tool they have, karma.
  - How to you keep karma from affecting site business?
  - Short of the benevolent dictator model, is there a better approach to
    encoding user value, participation or relevance?

Not an easy problem to solve as most communities have discovered. For the most part HN seems to regulate rather nicely. We are all guilty of veering off here and there but eventually things come back to a reasonable center. Maybe it's because posts age very quickly and it takes a lot to stay in the first page or two. Other types of sites don't have that advantage.

I don't see it as very HN-specific. It's the exact same pattern you find at Reddit and Slashdot, so if YC is mainly attempting to cultivate that property, looking for users with the highest Reddit karma might be a good strategy (and they've proven themselves to a larger audience, probably more representative of average consumers). The negative way of looking at it is that it's most important just to be in the general vicinity of activity and say something with either broad appeal, or provocation.

In academia this is becoming a common strategy with the rise of metrics. Find a hot area, publish a paper with a linkbaity title and prominent keyword placement, and wait for the citations to roll in. In business, it's the strategy of choosing your field of endeavor by picking the places where the most money is flowing around, and then just doing something average in that space.

That is interesting. If you want more information on them, including photos and twitter info, you can see it on http://www.hnwho.com

It depends on the topic, lots of popular legal based threads have had a very in depth reply from grellas with a ton of upvotes. I think in that situation although not everyone would fully grasp the content they are upvoting in appreciation that someone who is an expert in the field has taken the time to explain and simplify the complexities of a legal case or argument.

In a previous comment on karma systems I suggested a scheme where there was a voting 'diamond' up, down, left, right, and center. Where up was "good comment/submission" and down was "not-good comment/submission", left was "less like this", and right was "more like this" and center was perfect.

That allows you to establish both what you like the site to have in it, and how you much you liked this instance of it. Perfect would mean simply that "spot on" or "this."

The second part of the system would then adjust the presentation of your karma (to you and others) unscaled vector along the line to the 'perfect' site in the context of the viewer.

To use a simple example, lets say HN was over run with cat pictures, and I thought they were the best so I vote like this and "+1" (up and to the right), You the reader hate cat pictures so you always vote down/less like this.

So we've created virtual root nodes at +loves cat pictures and one at +hates cat pictures. Now I can plot your karma as negative with respect to mine if we assume that the center point between your root nexus and mine is 0. Now do this with enough topics and you get multiple consituencies all in the same discussion space where your view of their karma will inform you how likely it is you'll like to read what they wrote :-)