Qualifying "Web Karma": It shouldn't be a game.(unwieldy.net)
unwieldy.net
Qualifying "Web Karma": It shouldn't be a game.
http://www.unwieldy.net/inck/thoughts/qualifying-web-karma-it-shouldnt-be-a-game
11 comments
The whole karma discussion of the last couple of days has really given me a lot of stuff to think about. It has to be possible to 'crack' this problem somehow.
The problems that I've seen listed so far enumerated are:
- class hierarchy - groupthink - feedback loops - old hands vs newcomers (aka the incrowd problem)
I'll go back through all the stuff that was said on this subject and see if I can dig up more. It's a very interesting problem, possibly one of the most interesting ones that I have seen in a long time, it is worthy of a serious study.
The problems that I've seen listed so far enumerated are:
- class hierarchy - groupthink - feedback loops - old hands vs newcomers (aka the incrowd problem)
I'll go back through all the stuff that was said on this subject and see if I can dig up more. It's a very interesting problem, possibly one of the most interesting ones that I have seen in a long time, it is worthy of a serious study.
This is my take on karma: I don't care how much karma I have. I just make (hopefully) good comments. And if people disagree with me, it's okay, we're all entitled to an opinion.
But if I get really downmodded, I think about whether my comment was rude or something.
I don't need an orange username to feel good about myself.
But if I get really downmodded, I think about whether my comment was rude or something.
I don't need an orange username to feel good about myself.
I always loved this quote from Hannibal : "Will they give you a medal, Clarice, do you think? Would you have it professionally framed and hang it on your wall to look at and remind you of your courage and incorruptibility? All you would need for that, Clarice, is a mirror."
Self-esteem and intrinsic motivation are so much more powerful than extrinsic incentives. All these karma systems can do is, at best, reflect how people feel about themselves. The real challenge is to have a community that attracts the right people. Then, the karma system will show the quality of the community.
Self-esteem and intrinsic motivation are so much more powerful than extrinsic incentives. All these karma systems can do is, at best, reflect how people feel about themselves. The real challenge is to have a community that attracts the right people. Then, the karma system will show the quality of the community.
Exactly. I don't really need the karma for me, but we need it as a community to tell good and bad content apart.
Reposting a comment I made on the new comment features thread:
...I'd prefer having the system make its values explicit via descriptors for comments (think Slashdot or Plastic). My top two would be "insightful" and "uncharitable". The former encourages what's best about this community, the latter strongly discourages what's worst. A lot of deep and pointless threads are caused by commenters reacting to their own hasty interpretations of what someone said rather than even-handedly responding to them.
Also, no "Funny".
http://news.ycombinator.com/item?id=467281
I've also suggested (and agreed with later suggestions) that the karma score should be kept hidden. I'm not adverse to replacing "up/down" with "insightful/uncharitable" entirely.
...I'd prefer having the system make its values explicit via descriptors for comments (think Slashdot or Plastic). My top two would be "insightful" and "uncharitable". The former encourages what's best about this community, the latter strongly discourages what's worst. A lot of deep and pointless threads are caused by commenters reacting to their own hasty interpretations of what someone said rather than even-handedly responding to them.
Also, no "Funny".
http://news.ycombinator.com/item?id=467281
I've also suggested (and agreed with later suggestions) that the karma score should be kept hidden. I'm not adverse to replacing "up/down" with "insightful/uncharitable" entirely.
Last time this topic was topical on HN (about 6 months ago) I put my thoughts in a blog post http://www.shirlawsonline.com/blogs/199-my-pond-theory-or-ho...
I posed that size limits on membership and censorship were options - even using Bayesian filtering as a tool.
If I re-wrote those posts, I would definitely include descriptors as a partial solution. The karma game falls apart because the people whose thoughts are most encouraged are those who care about karma the least.
edit: I'm only applying this to social sites like those discussed here. On a site like eBay, there is much greater utility in the karma number and also much more restriction over its use and what it represents.
I posed that size limits on membership and censorship were options - even using Bayesian filtering as a tool.
If I re-wrote those posts, I would definitely include descriptors as a partial solution. The karma game falls apart because the people whose thoughts are most encouraged are those who care about karma the least.
edit: I'm only applying this to social sites like those discussed here. On a site like eBay, there is much greater utility in the karma number and also much more restriction over its use and what it represents.
great article.
slashdot is awesome. reddit and hn became popular because of the communities, despite the loss of technical goodness each time.
at the very least, i wish the quantifiable part of karma was kept strictly private.
slashdot is awesome. reddit and hn became popular because of the communities, despite the loss of technical goodness each time.
at the very least, i wish the quantifiable part of karma was kept strictly private.
But quantifiable karma and leaderboards that result from the addition of numerical values serve as incentives for users to contribute to the community.
For the site owner, 'the game' is in his/her interest because it pulls people in to participate. It gives recognition to the efforts of site users.
I'm not sure how effective 'good', 'very good' or 'poor' will perform as a substitute.
For the site owner, 'the game' is in his/her interest because it pulls people in to participate. It gives recognition to the efforts of site users.
I'm not sure how effective 'good', 'very good' or 'poor' will perform as a substitute.
I have not seen as many people with orange nicks posting lately. I had an orange name earlier today, but my high-scoring comments have fallen out of my recent history.
Are the orange users being very stringent about what they post? I don't blame them.
Are the orange users being very stringent about what they post? I don't blame them.
pg discovered on implementation that this was unnecessarily slowing the site. So it's been rewound (hence, those who qualified for the orange name are back at the generic grey, and likely still posting.)
http://news.ycombinator.com/item?id=468599
http://news.ycombinator.com/item?id=468599
Ah, indeed - thanks.
I'm pretty sure I already said this last time meta-talk maddness came up.
Yup, here it is. http://news.ycombinator.com/item?id=373992
Yup, here it is. http://news.ycombinator.com/item?id=373992
It sounds like pg just implemented something like this -- grayish-orange vs. gray -- a qualitative measure ...
The metric is slightly different, but the author basically got want he wanted.
The metric is slightly different, but the author basically got want he wanted.
I think with all the advances in scrip systems, social psychology, and game theory these days, there should be a way to quantify and award karma that incentivizes productive discussion.
karma has shortcomings, but of course this stuff has to be built, and that takes time and testing. i'm sure pg knew about the shortcomings of a basic karma system going into it, but he probably had a lot of other things to tend to before going to improve karma. it makes for a good conversational starter on the topic though.
The problem is that no one has yet invented a karma "game" that can't be manipulated. Still, when you're dealing with a large community the karma game is the only system that has shown promise. It would be great if just having clear guidelines was enough, but it doesn't work that way.