New Karma System (Theory)

[Uchiha Nick]

i for one dislike Karma, but i do have some ideas for it;

when voting, simply create a "why" comment. if its short, give a few extra points to the voter, if its long, more.

[MKruer]

i for one dislike Karma, but i do have some ideas for it;

when voting, simply create a "why" comment. if its short, give a few extra points to the voter, if its long, more.

That was something I was interested in. however in a slightly different fashion. In order to increase the karma, the post would have to be over a certain length. I know that too many people reply with simple one liners such as “cool idea” which currently if some one reply to the first, “thanks” the cool idea would get credit despite the fact that it was pointless reply. So yes their should be a minimum limit to qualify a post as having karma.

[DragonlordP]

Do you really think that administrators will waste their time checking out comments? Because I am 101% sure that after a while most comments will be like "aaaaaaaaaaaaaaaaaaaa" just to match the minimum length.

[vph]

well, to make it simple, every registered member should be able to rate a topic, regardless of weather he/she replies or not. Making certain technical requirements only complicates the manner, and more important imposing unnecessary constraints on users. Why? in order to make a rating, they MUST reply AND according to spec. That's too much as it goes into a realm of psychological uncertainty.

But these are details.

--------

Let's summarize the ideas (because I feel like some are repeating the old ideas here and for the sake of what's coming up next) that I think the original poster would agree: Someone's karma is affected in two ways (a) if his rating conforms with the norm; this encourages responsibility, and (b) when his posts are rated; this encourages quality.

--------

Now, I'd like to discuss the updating of karma in a manner that wasn't mentioned in the original theory.

Suppose A starts a topic, and B, C, D, E reply, and also rate A's topic.

Now the original theory says that A's karma should be updated according to the rating of B, C, D, and E, where each rating takes into account each of these individual's karma.

Up to now, the theory also says that, based on the collective, average rating of B, C, D, and E, the karma of each of B, C, D, and E is also affected based on how much each is conformed to that average rating.

However, .... MORE CAN BE SAID.

In practice, not only B rates A's topic, but likely B also rates the replies of C, D, and E. The same holds of each of C, D, and E.

This situation creates a complex scenerio, which however could help to compute karma more accurately.

For example, if C, D, and E each rates very highly B's reply to A's topic. Then, I'd argue that B's original rating (which is rated very highly by C,D, and E as said) to A's topic should be judged more in terms of weight. In fact, if C, D, and E unanimously think that B's reply to A is great, then B's rating to A should be affected by its rating from C, D, and E, as it is by B's own karma.

What this creates is a complete graph (or a clique) among all the participants of a topic, in this case A, B, C, D, and E. They rate each other's posts. And their decision should be have retrospective consequences. In other words, if after a long discussion with several rounds of rating, let's say everyone rates B's posts (within that thread) very highly, then B's original rating of A should be retrospectively counted more.

Then update mechanism is complicated but it makes the theory more reasonable, I think.

-------------


This mechanism, I think, will be more accurate, if people behave as they would without the karma system. If they try to abuse the system, I don't know what consequences it can have.

Then again, I think the karma system is not something that works for everyone.

[MKruer]

well, to make it simple, every registered member should be able to rate a topic, regardless of weather he/she replies or not. Making certain technical requirements only complicates the manner, and more important imposing unnecessary constraints on users. Why? in order to make a rating, they MUST reply AND according to spec. That's too much as it goes into a realm of psychological uncertainty.

Yes, but what I was trying to say is that in any registered users replies but that does not rated the post replied to would automatically get 5 (middle value) So by virtue of replying you are giving a middle ground credit to the first post. This assumes of course that the average post is neutral in value.

Let's summarize the ideas (because I feel like some are repeating the old ideas here and for the sake of what's coming up next) that I think the original poster would agree: Someone's karma is affected in two ways (a) if his rating conforms with the norm; this encourages responsibility, and (b) when his posts are rated; this encourages quality.

No problem here.

Now, I'd like to discuss the updating of karma in a manner that wasn't mentioned in the original theory.

Suppose A starts a topic, and B, C, D, E reply, and also rate A's topic.

Now the original theory says that A's karma should be updated according to the rating of B, C, D, and E, where each rating takes into account each of these individual's karma.

Up to now, the theory also says that, based on the collective, average rating of B, C, D, and E, the karma of each of B, C, D, and E is also affected based on how much each is conformed to that average rating.

However, .... MORE CAN BE SAID.

In practice, not only B rates A's topic, but likely B also rates the replies of C, D, and E. The same holds of each of C, D, and E.

This situation creates a complex scenerio, which however could help to compute karma more accurately.

For example, if C, D, and E each rates very highly B's reply to A's topic. Then, I'd argue that B's original rating (which is rated very highly by C,D, and E as said) to A's topic should be judged more in terms of weight. In fact, if C, D, and E unanimously think that B's reply to A is great, then B's rating to A should be affected by its rating from C, D, and E, as it is by B's own karma.

What this creates is a complete graph (or a clique) among all the participants of a topic, in this case A, B, C, D, and E. They rate each other's posts. And their decision should be have retrospective consequences. In other words, if after a long discussion with several rounds of rating, let's say everyone rates B's posts (within that thread) very highly, then B's original rating of A should be retrospectively counted more.

Then update mechanism is complicated but it makes the theory more reasonable, I think.

This sounds great, and if we could keep a history of votes ever for a while it could be done. But the question then goes down to how to calculate it?

This mechanism, I think, will be more accurate, if people behave as they would without the karma system. If they try to abuse the system, I don't know what consequences it can have.

This is what needs to be though out. To a great extent the normalization process will alleviate a great deal of these issues, but still needs to be worked on further. If you can come up with some type of equasion for you above idea then we can logically think it out and think of ways to abuse it and correct it.

Then again, I think the karma system is not something that works for everyone.

Hey its not mandatory, but any system developed should be developed with the “everyone” trying actively to abuse the system.

[MKruer]

Do you really think that administrators will waste their time checking out comments? Because I am 101% sure that after a while most comments will be like "aaaaaaaaaaaaaaaaaaaa" just to match the minimum length.

Im sure that there is a some way to create a somewhat automated veification system. that would work most of the time.

[vph]

This sounds great, and if we could keep a history of votes ever for a while it could be done. But the question then goes down to how to calculate it?

vph wrote:
This mechanism, I think, will be more accurate, if people behave as they would without the karma system. If they try to abuse the system, I don't know what consequences it can have.


This is what needs to be though out. To a great extent the normalization process will alleviate a great deal of these issues, but still needs to be worked on further. If you can come up with some type of equasion for you above idea then we can logically think it out and think of ways to abuse it and correct it.

Yes, but before getting dirty, down to the calculation level, I think the higher level of what affects what needs to be worked out first. The design of the theory if you will. After you have all of the requirements then we can talk about implementation.

Right now, are there other scenareos that should be looked at?

[Uchiha Nick]

hmz, no matter how you look at it, the system will be cheated. just dont use it.

[Mr.Jester]

hmz, no matter how you look at it, the system will be cheated. just dont use it.

I agree. No matter what law is created, it will be cheated. Lets not use laws.

[MKruer]

No system is ever going to be perfect, (perfect is a concept) However a system can be designed in such a way that it is virtually impossible to “cheat” it. That is the reason for this thread, to figure out the best system(s) to make it more difficult to cheat the system and maintain the integrity of the original Karma concept.