As a counterpoint, my highest rated comments are huge walls of text. This could be because (a) I don’t make a lot of jokes or (b) I make crappy jokes or (c) people like my walls of text more than the typical wall of text or (d) something else.
I keep an eye on my karma and have noticed these things that I believe are related to your post:
Talking about the karma system has fallen out of favor. I think people are getting tired of it.
Asking why something was downvoted usually brings more upvotes unless you really, really deserved the downvotes. This latter case will probably be swarmed with downvotes.
Some (many?) people vote with an end score in mind. These meta-voters have more affect on threshold comments that fluctuate between −2 and 2.
Long conversations will generally pull between −1 and +2 per post. Most of my karma comes from lengthy discussions in the comments. Even if the top level post was only rated +2 I will net almost 100 karma points from the post and comments. (Note: I haven’t actually added this up. It may be closer to 75 karma.)
Quick responses pointing out third alternatives or simple problems get upvoted and usually roam between +2 and +14. If you want to get your karma higher, this is the easiest way. Comment immediately after a post is submitted and point out the most obvious flaw respectfully and concisely. Don’t try to make a point, just note an error. If you get in before the rest of us, you will probably get +4 or higher.
Long responses pointing out serious problems get upvoted but have a lower chance of pulling large amounts of karma than quick responses. However, after the first wave of quick responses, only the longer comments are true candidates for higher karma. I suspect that a long response to a quick comment has a good chance to do well, but haven’t really watched those comments yet.
Jokes are upvoted when they are either extremely funny or solidly funny and on topic. Randomness is upvoted if it is an inside joke, otherwise stays around +0. Sarcasm is appreciated but has the problem of being mistaken for nonsarcasm.
Extending the point or conversation of a top-level post gets upvoted. Most of mine get between +2 and +4. Examples would be almost every comment I have made while reading the sequences.
Aggressiveness is generally poorly received on technical topics. Aggressiveness is easier to get away with when dealing with fuzzy topics. I attribute this mostly to margin of error. Technical topics are harder to be bulletproof. I am still having a hard time predicting which non-technical top level posts are voted up. I suspect this is because I don’t know what is already been discussed or that the issue is a technical topic that I have misidentified.
Self-depreciation is a huge karma pull. Both my highest rated post and comment were essentially me slamming myself over and over. Each were voted higher than +20.
Bullet point lists of extensions, ideas, questions, and so on seem to do about as well or better than long paragraphs of text. The walls of text may be harder to skim for goodies or the bullet point list better organized?
Non-aggressive requests for clarification or information are not generally downvoted. Mine seem to roam between 0 and +2 karma. If the question and response delve into a lengthy but friendly conversation, I seem to get between 0 and +2 karma for each of my comments. If a solid agreement or conclusion is reached, the capping comment gets about double whatever the individual comments were getting.
Posting nearby “famous” people amps up the karma action. Replying to EY, Alicorn, Vladimir_Nesov, pjeby, et al will increase the amount of people that read your comments. The reasons for this are varied. The four I used here are just names that popped into my head. Also, some people seem to vote up conversations they are in while others do not. A few downvote anyone disagreeing with them. The people that matter generally fit these criteria: (a) top contributer (b) easily recognizable name (c) frequent poster/commenter (d) abnormal amount of recent activity (e) holds atypical beliefs for the community (f) a troll.
Better grammar, spelling, and language increases the likelihood that your comment will move upward faster.
Comments quickly upvoted higher than typical seem to either (a) go through the roof or (b) get meta-voted back to between +1 and +3.
Telling people to vote in a particular way tends to produce easy to predict results but not in a manner that is easy to describe.
And oh wow did that get long. Do note that this is all being typed from the top of my head using myself and the comments I read as an example. Naturally, the above does not dictate how people vote.
EDIT: I guess for fun, I predict that this comment gets ROT13orgjrra cbfvgvir gjb naq cyhf fvk xneznROT13.
EDIT 2: I would normally downvote a post such as this but elsewhere in the comments you seemed to have received the message and was wondering about deleting it. So I just left it as it is. Also relevant: I generally do not upvote jokes unless they are truly amazing.
There also appears to be a non-linear effect where posts that reach a certain threshold of votes get even more votes—like there’s some kind of 80⁄20 power law at work. I can think of a number of possible explanations for that phenomenon but it would be interesting to see if the data backs up my impression first. I predict that the Gini coefficient for comment karma is > 0.3. Does anyone know a better way to obtain the data than a web scraper?
ETA: I’m referring to the Gini coefficient for individual comments’ karma rather than for individual commenters’.
ETA2: How to handle negative karma? I’d suggest ignoring negative scoring posts but other suggestions welcome.
This analysis looks brilliant, but the impression is sort of spoiled by the fact that the prediction of karma gain failed. It is even close to self-referential paradoxes: comment reading “This comment will get X karma” should be upvoted because of its precise prediction when it actually has X karma, but upvoting makes the precision worse and may attract later downvotes.
Posting nearby “famous” people amps up the karma action
This in itself strikes me as a sufficient incentive to aim at getting into the “Top contributors” list. My “return on time investement” for participating in LW is what I learn (often tangentially) from getting into interesting conversations. The lower my chance of being ignored, the higher my expected utility from any contribution to LW.
There is a slight difference between being a top contributor and being famous as I am mentioning it here.
My current karma experiment is deliberately not posting comments I think are worth less than 2 karma unless I have a compelling reason to do so (such as asking for help or information). My goal is to increase the quality of my comments to the point that someone could think, “What has MrHen posted recently?” and the answer is more impressive than a series of one-liners and nitpicky comments.
Ideally, this will increase the weight of my words to the point that when I speak, people will listen. It is a pure, straightforward status grab, but one that doesn’t involve gobbling up karma. The pinnacle of the status tree at LessWrong is to have someone think, “I, the reader, am wrong” instead of, “they, the writer, are wrong.”
I am compiling a mental list for LW status games, rewards, and penalties that is similar to the karma list above. I am not much for status games but the gaming here seems to be harder to avoid than in real life. (“Avoid” is the wrong term but conveys the right intent. Status games are hard to “avoid.”)
My biggest regret with the above karma list is that I have no good way to verify or catalogue my comments and their karma. LW just doesn’t have any tools to make such a thing easy. I worry that the status list will be even further removed from reality—possibly to the point of being unable to predict anything at all.
In any case, I like to think that I am getting better in regards to comment quality and topical knowledge. My karma rating keeps going up, so that’s a good sign.
EDIT: Replaced “I think will be rated lower than 2 karma” with “I think are worth less than 2 karma.”
Not so much ideas as things that I think would be helpful but will be buried in obscurity at the level I am.
That being said, this is helped more by learning the basics through reading the sequences than playing status games. My thoughts on status should be taken with the clarification that I am primarily seeking to learn and am thinking about status because I find it interesting.
If I can take my comments and change my behavior so as to be looking forward and increase my status without hampering my ability to learn… why shouldn’t I? When I look at the various status levels here at LessWrong I notice the highest level I can see, which is the one I mentioned in my previous comment.
These comments are meant to be taken as observations for people who are curious or interested. I am not nearly as concerned with status as these comments may imply. I just know status is there and wonder about it in the same way I wonder about most things.
My biggest regret with the above karma list is that I have no good way to verify or catalogue my comments and their karma. LW just doesn’t have any tools to make such a thing easy.
I think the format itself (a list, where each item is at least a whole sentence) has a connotation of thoroughness and authority; and it reduces the TL;DR effect by presenting the content in small-but-not-trivial separable pieces.
Also, I see a “wall of text” as not having bullet points: being a series of long paragraphs without much structure visible without reading the text itself. The term is being misapplied to this comment in my opinion.
As a counterpoint, my highest rated comments are huge walls of text. This could be because (a) I don’t make a lot of jokes or (b) I make crappy jokes or (c) people like my walls of text more than the typical wall of text or (d) something else.
I keep an eye on my karma and have noticed these things that I believe are related to your post:
Talking about the karma system has fallen out of favor. I think people are getting tired of it.
Asking why something was downvoted usually brings more upvotes unless you really, really deserved the downvotes. This latter case will probably be swarmed with downvotes.
Some (many?) people vote with an end score in mind. These meta-voters have more affect on threshold comments that fluctuate between −2 and 2.
Long conversations will generally pull between −1 and +2 per post. Most of my karma comes from lengthy discussions in the comments. Even if the top level post was only rated +2 I will net almost 100 karma points from the post and comments. (Note: I haven’t actually added this up. It may be closer to 75 karma.)
Quick responses pointing out third alternatives or simple problems get upvoted and usually roam between +2 and +14. If you want to get your karma higher, this is the easiest way. Comment immediately after a post is submitted and point out the most obvious flaw respectfully and concisely. Don’t try to make a point, just note an error. If you get in before the rest of us, you will probably get +4 or higher.
Long responses pointing out serious problems get upvoted but have a lower chance of pulling large amounts of karma than quick responses. However, after the first wave of quick responses, only the longer comments are true candidates for higher karma. I suspect that a long response to a quick comment has a good chance to do well, but haven’t really watched those comments yet.
Jokes are upvoted when they are either extremely funny or solidly funny and on topic. Randomness is upvoted if it is an inside joke, otherwise stays around +0. Sarcasm is appreciated but has the problem of being mistaken for nonsarcasm.
Extending the point or conversation of a top-level post gets upvoted. Most of mine get between +2 and +4. Examples would be almost every comment I have made while reading the sequences.
Aggressiveness is generally poorly received on technical topics. Aggressiveness is easier to get away with when dealing with fuzzy topics. I attribute this mostly to margin of error. Technical topics are harder to be bulletproof. I am still having a hard time predicting which non-technical top level posts are voted up. I suspect this is because I don’t know what is already been discussed or that the issue is a technical topic that I have misidentified.
Self-depreciation is a huge karma pull. Both my highest rated post and comment were essentially me slamming myself over and over. Each were voted higher than +20.
Bullet point lists of extensions, ideas, questions, and so on seem to do about as well or better than long paragraphs of text. The walls of text may be harder to skim for goodies or the bullet point list better organized?
Non-aggressive requests for clarification or information are not generally downvoted. Mine seem to roam between 0 and +2 karma. If the question and response delve into a lengthy but friendly conversation, I seem to get between 0 and +2 karma for each of my comments. If a solid agreement or conclusion is reached, the capping comment gets about double whatever the individual comments were getting.
Posting nearby “famous” people amps up the karma action. Replying to EY, Alicorn, Vladimir_Nesov, pjeby, et al will increase the amount of people that read your comments. The reasons for this are varied. The four I used here are just names that popped into my head. Also, some people seem to vote up conversations they are in while others do not. A few downvote anyone disagreeing with them. The people that matter generally fit these criteria: (a) top contributer (b) easily recognizable name (c) frequent poster/commenter (d) abnormal amount of recent activity (e) holds atypical beliefs for the community (f) a troll.
Better grammar, spelling, and language increases the likelihood that your comment will move upward faster.
Comments quickly upvoted higher than typical seem to either (a) go through the roof or (b) get meta-voted back to between +1 and +3.
Telling people to vote in a particular way tends to produce easy to predict results but not in a manner that is easy to describe.
And oh wow did that get long. Do note that this is all being typed from the top of my head using myself and the comments I read as an example. Naturally, the above does not dictate how people vote.
EDIT: I guess for fun, I predict that this comment gets ROT13orgjrra cbfvgvir gjb naq cyhf fvk xneznROT13.
EDIT 2: I would normally downvote a post such as this but elsewhere in the comments you seemed to have received the message and was wondering about deleting it. So I just left it as it is. Also relevant: I generally do not upvote jokes unless they are truly amazing.
There also appears to be a non-linear effect where posts that reach a certain threshold of votes get even more votes—like there’s some kind of 80⁄20 power law at work. I can think of a number of possible explanations for that phenomenon but it would be interesting to see if the data backs up my impression first. I predict that the Gini coefficient for comment karma is > 0.3. Does anyone know a better way to obtain the data than a web scraper?
ETA: I’m referring to the Gini coefficient for individual comments’ karma rather than for individual commenters’.
ETA2: How to handle negative karma? I’d suggest ignoring negative scoring posts but other suggestions welcome.
This analysis looks brilliant, but the impression is sort of spoiled by the fact that the prediction of karma gain failed. It is even close to self-referential paradoxes: comment reading “This comment will get X karma” should be upvoted because of its precise prediction when it actually has X karma, but upvoting makes the precision worse and may attract later downvotes.
Yeah, I was way off. I didn’t think people would be that interested in karma theory. I think the big oops was the first bullet point.
This in itself strikes me as a sufficient incentive to aim at getting into the “Top contributors” list. My “return on time investement” for participating in LW is what I learn (often tangentially) from getting into interesting conversations. The lower my chance of being ignored, the higher my expected utility from any contribution to LW.
There is a slight difference between being a top contributor and being famous as I am mentioning it here.
My current karma experiment is deliberately not posting comments I think are worth less than 2 karma unless I have a compelling reason to do so (such as asking for help or information). My goal is to increase the quality of my comments to the point that someone could think, “What has MrHen posted recently?” and the answer is more impressive than a series of one-liners and nitpicky comments.
Ideally, this will increase the weight of my words to the point that when I speak, people will listen. It is a pure, straightforward status grab, but one that doesn’t involve gobbling up karma. The pinnacle of the status tree at LessWrong is to have someone think, “I, the reader, am wrong” instead of, “they, the writer, are wrong.”
I am compiling a mental list for LW status games, rewards, and penalties that is similar to the karma list above. I am not much for status games but the gaming here seems to be harder to avoid than in real life. (“Avoid” is the wrong term but conveys the right intent. Status games are hard to “avoid.”)
My biggest regret with the above karma list is that I have no good way to verify or catalogue my comments and their karma. LW just doesn’t have any tools to make such a thing easy. I worry that the status list will be even further removed from reality—possibly to the point of being unable to predict anything at all.
In any case, I like to think that I am getting better in regards to comment quality and topical knowledge. My karma rating keeps going up, so that’s a good sign.
EDIT: Replaced “I think will be rated lower than 2 karma” with “I think are worth less than 2 karma.”
Why do you want more status?
Do you have ideas that you think are good but won’t be heard unless you’re more respected?
Not so much ideas as things that I think would be helpful but will be buried in obscurity at the level I am.
That being said, this is helped more by learning the basics through reading the sequences than playing status games. My thoughts on status should be taken with the clarification that I am primarily seeking to learn and am thinking about status because I find it interesting.
If I can take my comments and change my behavior so as to be looking forward and increase my status without hampering my ability to learn… why shouldn’t I? When I look at the various status levels here at LessWrong I notice the highest level I can see, which is the one I mentioned in my previous comment.
These comments are meant to be taken as observations for people who are curious or interested. I am not nearly as concerned with status as these comments may imply. I just know status is there and wonder about it in the same way I wonder about most things.
Have you tried Wei Dai’s script?
You wrote this (now high-rated) wall of text just to further prove your point, didn’t you?
This reminds me of that my highest rated so far answer at Stack Overflow was a big bullet pointed list.
I think the format itself (a list, where each item is at least a whole sentence) has a connotation of thoroughness and authority; and it reduces the TL;DR effect by presenting the content in small-but-not-trivial separable pieces.
Also, I see a “wall of text” as not having bullet points: being a series of long paragraphs without much structure visible without reading the text itself. The term is being misapplied to this comment in my opinion.
I agree that a bullet point list isn’t generally considered a wall of text.