Yep, that’s my experience as well. Recently, I decided “screw what LW thinks” and started posting more thoughts of mine, and they’re all getting upvoted. My vague intuitions about how many upvotes my posts will get doesn’t seem to correlate very well with how many upvotes they actually get either. This is probably true for other people as well.
The only potential problem with this, IMO, is if people think I’m more of an authoritative source than I actually am. I’m just sharing random thoughts I have; I don’t do scholarly work like gwern.
It seems to me that there are lots and lots of people who want to write posts but they’re concerned about whether those posts will be received well. I’ve read, also, that more people put “public speaking” as their worst fear than “death” when surveyed. If we made a karma prediction tool, maybe that would help get people posting here. Here’s what I’m thinking:
First, we could create a checklist of the traits that we think will get a LessWrong post upvoted. For instance:
Is there an obvious main point or constructive goal?
Is the main point supported / is there a reasonable plan for the constructive goal? (Or are they otherwise framed in the correct context “This is hypothetical” or whatever.)
What type of support is included (math, citations, graphics, etc).
Was the topic already covered?
Is it a topic of interest to LessWrong?
Is it uplifting or unhappy?
(Or do a separate survey that asks people’s reasons for upvoting / downvoting and populate the checklist with those.)
Then we could post the checklist as a poll in each new post and article for a while.
Then we could correlate the karma data with the checklist poll data and test it to see how accurately it predicts a post’s karma.
If you had a karma prediction tool, would it help you post more?
[pollid:413]
Posting that checklist as a poll in each new post would likely end up irritating people.
A simpler approach, with the twin advantages of being simpler and being something one can do unilaterally, would be to just count the proportion of recent, non-meetup-related Discussion posts with positive karma. Then you could give potential post authors an encouraging reference class forecast like “85% of non-meetup Discussion posts get positive karma”.
You know what? That is simple and elegant. I like that about it… but in the worst case scenario, that will encourage people to post stuff without thinking about it because they’ll make the hasty generalization that “All non-meetup posts have an 85% chance of getting some karma” and even in the best case scenario, a lot of people will probably be thinking something along the lines of “Just because Yvain and Gwern and people who are really good at this get positive karma doesn’t mean that I will.”
Yep, that’s my experience as well. Recently, I decided “screw what LW thinks” and started posting more thoughts of mine, and they’re all getting upvoted. My vague intuitions about how many upvotes my posts will get doesn’t seem to correlate very well with how many upvotes they actually get either. This is probably true for other people as well.
The only potential problem with this, IMO, is if people think I’m more of an authoritative source than I actually am. I’m just sharing random thoughts I have; I don’t do scholarly work like gwern.
It seems to me that there are lots and lots of people who want to write posts but they’re concerned about whether those posts will be received well. I’ve read, also, that more people put “public speaking” as their worst fear than “death” when surveyed. If we made a karma prediction tool, maybe that would help get people posting here. Here’s what I’m thinking:
First, we could create a checklist of the traits that we think will get a LessWrong post upvoted. For instance:
Is there an obvious main point or constructive goal?
Is the main point supported / is there a reasonable plan for the constructive goal? (Or are they otherwise framed in the correct context “This is hypothetical” or whatever.)
What type of support is included (math, citations, graphics, etc).
Was the topic already covered?
Is it a topic of interest to LessWrong?
Is it uplifting or unhappy?
(Or do a separate survey that asks people’s reasons for upvoting / downvoting and populate the checklist with those.)
Then we could post the checklist as a poll in each new post and article for a while.
Then we could correlate the karma data with the checklist poll data and test it to see how accurately it predicts a post’s karma.
If you had a karma prediction tool, would it help you post more? [pollid:413]
Posting that checklist as a poll in each new post would likely end up irritating people.
A simpler approach, with the twin advantages of being simpler and being something one can do unilaterally, would be to just count the proportion of recent, non-meetup-related Discussion posts with positive karma. Then you could give potential post authors an encouraging reference class forecast like “85% of non-meetup Discussion posts get positive karma”.
You know what? That is simple and elegant. I like that about it… but in the worst case scenario, that will encourage people to post stuff without thinking about it because they’ll make the hasty generalization that “All non-meetup posts have an 85% chance of getting some karma” and even in the best case scenario, a lot of people will probably be thinking something along the lines of “Just because Yvain and Gwern and people who are really good at this get positive karma doesn’t mean that I will.”
Unfortunately, I think it would be ineffective.
Fair points.