I’m replying to this even though it was retracted, because it’s an idea that I think needs to be discussed and I appreciate that you brought it up.
That said, I’m deeply skeptical that such a system would lead to less abuse, rather than more. I feel that karma systems should accurately convey information that is useful to the users, so I tend to be pretty touchy when I sense that they’re being abused. Statements like “they’re just meaningless numbers” indicates, to me, that a karma system is not performing its job correctly. Ideally, karma should convey several pieces of information, both to the audience and to the speaker:
How the audience feels about the speaker’s specific comment (this is the most important)
Why the audience feels that way about the speaker’s specific comment (this is almost as important as 1)
How the audience feels about the speaker in general (this is the most problematic piece of information, because all of the primate pack dynamic instincts come rushing in)
What the speaker should work on to improve themselves (this is very difficult to convey, because all the problems that cause 3 to fail are amplified when applied to 4).
I retracted it primarily because I read more of the thread and realized that the idea was too similar to another discussion which had already been brought up, which made the initial comment out of place and other parts repetitive. At that point, I was left with a “Since this has been brought up before, rather than posting it as a new idea, let me just retract it and upvote the other person who suggested it.”
But that being said, I’m not sure I fully understand your abuse concerns. If someone wanted to abuse the system, they can just sockpuppet up their karma without any other people. Using this method is substantially less prone to abuse since there is another person involved, and it seems to parallel things which would not be considered abusive. If you are asking for too much positive karma, presumably the other person would say so. (It does cost them a small amount of time/effort to karma repair you, after all.)
However, I suppose if you wanted to be completely sure you were not abusing the system, you could just ask your friend “Could you read all of my recent posts and let me know your thoughts?” If your friend saw “Wait, these are all at −1, but they’re all perfectly fine.” they would probably systematically upvote you even if you didn’t actually ask for upvotes. Asking for a friend to look at your work by itself does not seem like it could be considered abusive under any reasonable set of rules.
You could always ask for two entirely separate friends to BOTH review your work and have them not know who the other friend is (or for that matter, even that there WAS another friend reviewing your work, if needed.) They wouldn’t be able to easily form an echo chamber because they would not be in communication with each other.
Alternatively, you could, if you really wanted to avoid an echo chamber, explicitly ask Poster A “Can you review this with the intent of finding flaws? Crocker’s rules apply, be as harsh as you like.” and Poster B “Can you review this with the intent of finding good points?” It does not seem like that approach could form anything remotely like what I would consider an echo chamber.
That’s asking a fair amount from people I only know over the Internet. I could count on one hand the number of people that I expect a >50% chance of response (even refusal) if I asked them for this kind of help on this site.
That’s asking a fair amount from people I only know over the Internet. I could count on one hand the number of people that I expect a >50% chance of response (even refusal) if I asked them for this kind of help on this site.
I, or another person, would appear to be volunteering reading and offering relatively unskilled commentary on short blurbs of rationality related text. I know I personally already do that for fun when I post here, I just somewhat randomly select which items to reply to. This would essentially be what I would already be doing except people would be sending me what to read, so it doesn’t seem to be a lot of work for me at all, unless the response was unexpectedly massive.
I’m replying to this even though it was retracted, because it’s an idea that I think needs to be discussed and I appreciate that you brought it up.
That said, I’m deeply skeptical that such a system would lead to less abuse, rather than more. I feel that karma systems should accurately convey information that is useful to the users, so I tend to be pretty touchy when I sense that they’re being abused. Statements like “they’re just meaningless numbers” indicates, to me, that a karma system is not performing its job correctly. Ideally, karma should convey several pieces of information, both to the audience and to the speaker:
How the audience feels about the speaker’s specific comment (this is the most important)
Why the audience feels that way about the speaker’s specific comment (this is almost as important as 1)
How the audience feels about the speaker in general (this is the most problematic piece of information, because all of the primate pack dynamic instincts come rushing in)
What the speaker should work on to improve themselves (this is very difficult to convey, because all the problems that cause 3 to fail are amplified when applied to 4).
I retracted it primarily because I read more of the thread and realized that the idea was too similar to another discussion which had already been brought up, which made the initial comment out of place and other parts repetitive. At that point, I was left with a “Since this has been brought up before, rather than posting it as a new idea, let me just retract it and upvote the other person who suggested it.”
But that being said, I’m not sure I fully understand your abuse concerns. If someone wanted to abuse the system, they can just sockpuppet up their karma without any other people. Using this method is substantially less prone to abuse since there is another person involved, and it seems to parallel things which would not be considered abusive. If you are asking for too much positive karma, presumably the other person would say so. (It does cost them a small amount of time/effort to karma repair you, after all.)
However, I suppose if you wanted to be completely sure you were not abusing the system, you could just ask your friend “Could you read all of my recent posts and let me know your thoughts?” If your friend saw “Wait, these are all at −1, but they’re all perfectly fine.” they would probably systematically upvote you even if you didn’t actually ask for upvotes. Asking for a friend to look at your work by itself does not seem like it could be considered abusive under any reasonable set of rules.
What do you think of that alternative?
It seems better, but does seem to lead to a risk of echo-chamber situations.
Being human is hard.
You could always ask for two entirely separate friends to BOTH review your work and have them not know who the other friend is (or for that matter, even that there WAS another friend reviewing your work, if needed.) They wouldn’t be able to easily form an echo chamber because they would not be in communication with each other.
Alternatively, you could, if you really wanted to avoid an echo chamber, explicitly ask Poster A “Can you review this with the intent of finding flaws? Crocker’s rules apply, be as harsh as you like.” and Poster B “Can you review this with the intent of finding good points?” It does not seem like that approach could form anything remotely like what I would consider an echo chamber.
That’s asking a fair amount from people I only know over the Internet. I could count on one hand the number of people that I expect a >50% chance of response (even refusal) if I asked them for this kind of help on this site.
And I’m not sure anything is wrong with that.
I, or another person, would appear to be volunteering reading and offering relatively unskilled commentary on short blurbs of rationality related text. I know I personally already do that for fun when I post here, I just somewhat randomly select which items to reply to. This would essentially be what I would already be doing except people would be sending me what to read, so it doesn’t seem to be a lot of work for me at all, unless the response was unexpectedly massive.
I guess what I am considering is like Jsalvatier’s help desk at http://lesswrong.com/lw/eto/lesswrong_help_desk_free_paper_downloads_and_more/ but at a lower level for people who have simple questions like “Why was this downvoted?”
Can you go into more detail?