Promotion should be based on whether an article would be valuable to random first time visitors from the wider Internet. These articles should be +5 Informative, at least somewhat relevant to the ordinary life, and readable out of context.
Are You Anosognosic is a question the community. A random visitor must read several articles to know why are you asking that question, and what is it all about.
Incidentally, Sayeth the Girl shouldn't be promoted regardless of its score because it's meta. Random visitors are unlikely to benefit from it.
Promotion policies need to be informed by how the site is actually used.
Eliezer, you seem to have trouble distinguishing normative and positive beliefs about usage of this site. If you told us how we should use it, maybe we'd do it. Or maybe we'd tell you it's not going to happen. But there's little point in having secret beliefs about what we should do. This problem is clear to me in discussions of the wiki, with people continually expressing surprise about how you wanted it used. Maybe this problem doesn't exist at all here, but the wiki story suggests that it does. I'm pretty sure you have different beliefs from CannibalSmith (who has normative beliefs and might not realize that they aren't shared), but maybe CS is an outlier.
Like CannibalSmith, I don't use the front page. Newcomers do use the front page. My impression is that commenting is finished on a fair number of articles before promotion, which suggests to me that most regulars don't care that much about promotion (though I imagine they do use the front page).
My usage also seems relevant to my voting. Like the front page, "hot posts" doesn't get my attention; instead I use new/recent posts. This reduces the salience of the score and I usually don't think to vote. But I do vote on date-ordered comments, so I'm confused.
It seems most submitted articles are promoted; six of the last ten recent posts are on the front page. That ranges from an article with a score of 8 to one with a score of 38. Furthermore, posting is relatively slow (the tenth most recent submitted is four days old). Because of this low rate, good articles do not slide under the radar. Consequently, voting does not appear necessary to separate the wheat from the chaff. I think the only way to improve the value of voting would be to both increase posting while lowering the bar in terms of quality. To use a simple analogy, there is more incentive to pick the good apples out of a barrel when you know there are worms in it. Currently there are no negative consequences to not voting articles up or down... especially when there are so few that you, Eliezer, are probably reading them all. If there were twenty articles a day, though, we would vote up a good one when we found it.
I'm not suggesting we must submit more articles or lower our expectations - I just think that's the primary reason higher volume sites like Reddit or Digg have higher rates of user voting. Same reason there is more voting within article comments on LW: there are more comments, and your vote may actually encourage people to read or ignore one.
i totally second this!!! Eliezer posts very few articles. I usually come back from work and sit down to read on LW for 2-3 hours. I was fighting myself and telling myself to make a bit of effort to vote, as it is helpful as pointed out by eliezer. But, bcos the number of posts is so less, it really doesn't matter. Eliezer can almost assume that we like the post, unless theres something negative about it written in the comments section. As pointed out by eirenicon, comment votes are very important and they certainly should be given more attention. yesterday i was reading the "alcohol thread" and it really helped to skim thru few and get past 250 comments!!!!
But if no one votes, I tend to assume the article is regarded by the readers as chaff. The four of ten that are not on the front page have relatively fewer votes.
True -- but this is not an instance of no one voting (I assume "Are You Anosognosic?" has received some downvotes as well because it is under 'Controversial'). The purpose of voting is to improve the signal to noise ratio, but there is so little noise on LW that voting is really only distinguishing between good signal and great signal. Nobody is afraid that not voting will diminish the current signal to noise ratio, are they?
I would be more likely to vote if there were more options. This isn't necessarily because I want others to have more information, but because voting is also a signal to myself about what I value (for enjoyment, education, and of course signaling).
At the very least, a "vote neutral" or "vote very up" would help a lot to distinguish the ok posts from the excellent ones that should be promoted.
Rather ironic that a site dedicated to rationality should depend to a large extent on, from what I can tell, is an irrational impulse to vote, no?
After all, what is the utility of voting, unless it is believed that everyone will act in the same way---but group behavior is not something you can control.
I suppose it could be useful to remind yourself of the more interesting comments and commenters, at least, as you suggest.
Oh my, in a fit of irrationality, I just voted up your comment.
You could make voting a post mandatory to comment on that post, so to submit a comment you get prompted to vote it up or down (or maybe neutral)
Or maybe just by having the vote up/down/neutral buttons next to the comment submit button, right in peoples faces, would make them more likely to vote.
I would vote on far more articles and comments if the feeds contained vote buttons, as I do not usually find any single item worthy of the additional effort of opening the web page.
Further, I would vote on more comments if the comment feeds contained context; at least the text of the comment's immediate ancestor.
What percentage of posts that you read do you upvote and downvote? I think my breakdown for posts has been something like this:
And for comments:
Speaking only for myself, "Are you Anosognosic" seemed to offer little that wasn't already covered in "The Strangest Thing An AI Could Tell You" or other posts and the comments thereon. I also upvote sparingly, though, so take with a grain of salt.
I agree with Eirenicon.
And for reference, I downvoted "Are You Anosognosic"; if more people did that, then that might make it seem like fewer people were voting on it. So far, your system seems fine, and it's probably not worth stressing over if there are not a lot of articles through which to sift.
Except in rare cases (like Wei Dai's Fair Division of Black-Hole Negentropy) I'm still using article upvotes to partially determine whether to promote articles to the front page - some informal mixture of "number of upvotes" + "editor's judgment". I mention this because while comment voting is still healthy, the amount of article voting seems to be dropping off. As of now I'm still drawing the inference that no one thinks "Are You Anosognosic?" worthy of promotion, or wants to see similar articles from me in the future - since other articles have at least gotten more votes than 0. But as the amount of article voting diminishes, it becomes harder to trust such inferences. Maybe people liked that article (or others I haven't promoted) and just didn't bother to upvote.
I'm posting this observation just in case people figure that upvoting articles doesn't make a difference. It does. It also encourages authors to write similar posts in the future, or alternatively not.