Near the beginning of this year Wei Dai asked why certain people don't post to LessWrong more often, and Yvain replied that:

Less Wrong requires no politics / minimal humor / definitely unambiguously rationality-relevant / careful referencing / airtight reasoning (as opposed to a sketch of something which isn't exactly true but points to the truth.) This makes writing for Less Wrong a chore as opposed to an enjoyable pastime.

But Kaj disagreed that this was the actual standard:

I agree with the "no politics" bit, but I don't think the rest are correct. I've certainly had "sketch of something that isn't quite true but points in the right direction" posts with no references and unclear connections to rationality promoted before (example), as well as ones plastered with unnecessary jokes (example).

This raises two questions: what is the real standard, and what should the standard be?

Because on the one hand, it's not clear Yvain is right, but on the other hand if he is right on the factual question, that standard seems way too high to me. It would suggest that, as John Maxwell says in the same thread, "The overwhelming LW moderation focus seems to be on stifling bad content. There's very little in place to encourage good content."

The wiki sort-of answers the factual question:

These traditionally go in Discussion:

  • a link with minimal commentary 
  • a question or brainstorming opportunity for the Less Wrong community

Beyond that, here are some factors that suggest you should post in Main:

  • Your post discusses core Less Wrong topics. 
  • The material in your post seems especially important or useful. 
  • You put a lot of thought or effort into your post. (Citing studies, making diagrams, and agonizing over wording are good indicators of this.) 
  • Your post is long or deals with difficult concepts. (If a post is in Main, readers know that it may take some effort to understand.) 
  • You've searched the Less Wrong archives, and you're pretty sure that you're saying something new and non-obvious.

But this isn't an entirely unambiguous answer: how many of the five "factors" does a post need to be in Main? Furthermore, it often seems that the "real" rules are significantly different than what the wiki says. Yvain's perception may be incorrect, but I think there were reasons why he (and presumably the people who upvoted his comment) had that perception. Also, Eliezer recently explained that:

Whenever a non-meta post stays under 5, I always feel free to move it to Discussion, especially if an upvoted comment has also suggested it. I don't always, but often do.

This makes me wonder what other poorly-publicized rules there are in this vicinity.

As for what the rules should be, I'm going to limit myself to two general suggestions:

  • The standard for posting in Main should not be so high that it makes posting at LessWrong feel like a chore, thereby chasing away good contributors like Yvain.
  • The standard should not be so high that it would force any significant portion of Eliezer's original sequences off into Discussion.

Finally, whatever standard we settle on, I think it's really important that we make it clearer to people what it is. Aside from the obvious benefits of doing that, I've found that trying to navigate the unclear Main/Discussion distinction is itself often enough to make blogging at LessWrong feel like a chore.

Edited to add: In terms of karma I'm currently the top contributor for the past 30 days on LessWrong by a wide margin. I managed this in spite of the fact that I'm in the middle of doing App Academy and have no time (this past week has been an exception because vacation). I take this not as evidence of how awesome I am, but as evidence that way too little quality content is being posted in Main. 

New Comment
103 comments, sorted by Click to highlight new comments since: Today at 10:29 AM
Some comments are truncated due to high volume. (⌘F to expand all)Change truncation settings

For me, promoted posts on Main are what should be our "public facade" blog, things worth reading for outsiders who don't have much time. Putting a post in Main instead of Discussion is just a clunky way of saying it should be promoted.

I don't know how much more attention promoted posts actually get, and I'm not sure it's worth the trouble (they may even get less attention because of regulars who check discussion but not main because stuff is rarely posted in main). The bastard category of "in main but not promoted" doesn't have a reason to exist (except maybe for Karma reasons?).

promoted posts on Main are what should be our "public facade" blog, things worth reading for outsiders who don't have much time.

This is the most meaningful distinction. It answers the Why of Main. All the other rules are just heuristics for How.

When we forget the purpose, the debate goes astray towards: "make it, uhm, what it is now, just a bit different, you know, better".


The thing I don't get is why LessWrong defines itself as a "community blog" but then—unlike every other blog I have ever read—presents visitors with a (basically) static page when they go to If this is a consular ship, then where is the ambassador? To get to the actual blog you're supposed to click Main, which is just weird. If is the 'main' thing then why isn't it the thing I first see when I visit

because main posts are often either meetup announcements or really indepth decision theory stuff that is both boring and opaque to newcomers. We don't get as many instantly exciting articles like a lot of the sequence posts as we used to
On the main page, the four most recent promoted articles are visible (as titles) on the left side, underneath the welcome. On my 1920x1080 monitor, the title of that section is partially visible on the bottom of the page. (This seems like a problem.)

I think that the Main/Discussion distinction is obsolete, and should be done away with. I think that there are a lot of neat ways to present posts to people, and that we could use the tagging system more effectively, and that meetups shouldn't be in Discussion, and so on.

I also think that this will not change without devoted programmer-hours, and that it may be better to do a fundraising campaign for programmer-time for new features than discuss policy changes.

Before searching the programmer time it would be good to have a statement from the powers that be that they would approve of implementing a particular change.

The distinction has confused me from the beginning. There is no clear workflow. At first, when I decided whether to post to Main or Discussion, it seemed that I had to choose based on ego. I ended up just posting everything to Discussion to let the powers-that-be promote it if they want to.

Why not just say that everything goes initially to Discussion, and the better items are boosted to Main?

Also, it was not clear to me how things are moved from Discussion to Main. As far as I can tell, Eliezer and a small group of anonymous admins decide. That's quite reasonable, but this policy should be clearly stated.

1KnaveOfAllTrades10y might be the answer, but yeah, clarity would be nice. It's possible that this is not done in practice because the moderators can't ensure that they'll read every article and be able to judge whether it stays or goes to main, and taking the time/coordinating to ensure every post is judged is not considered a better option.
I don't think moderator privileges are sufficient to promote or move posts: or if they are, I've missed the link on the interface. I can only delete comments and posts, not affect where they are. (Besides moving my own posts, of course.)
Ah yeah. I checked and I think I was thinking of the similar list of editors.
There are three categories, the list of moderators, the list of editors, and people with Editor on their about page, including Eliezer, wmoore, and matt, the latter two of whom are from Trike and probably don't promote posts. So there are two uses of the word "editor" and I don't trust any source that doesn't explicitly acknowledge that, even if it explicitly points to a particular meaning. ---------------------------------------- more detail in my original comment: Your link does not appear to me to be a list. After following it for a while, I found a list of "editors" distinct from "moderators," but am not convinced that it answers the question. Your link does have a prominent link, which it implies is a link to a list. But that link is also not a list, but a question asking for a list; the first comment contains links to lists of "editors" and "moderators." However, one of those editors denies that "editors" have the power to promote articles, so I don't think it really answers the question, even if your original link claims that they do. (Another distinction is that moderators delete comments and editors edit posts.) Also, people listed as editors are not the same people with the word Editor near their karma circle. In particular, Vladimir is listed but not labeled, while Eliezer is the opposite. In fact, no one on the editor list has the editor label.

As I suggested before, Main should be automatically generated based on post karma (e.g. 20+ karma posts should be auto-promoted). Sort of like Reddit front page. There is little reason to have two sections. No need to worry about suitability for Main ever again. The only downside I can think of is that Eliezer and others apparently only subscribe to Main, so they will not see new quality posts right away.

Upvotes don't work as a sole measure because easy content rises faster - just look at what happens in reddit. Even in smaller sub-reddits, top content is never best content.

Weight votes based on who voted. The simplest would be to just multiply an upvote (or downvote) by the voter's karma score. That would make a few users have a nearly overwhelming vote, so maybe weight by the log of the karma or another suitable function. Or, since this is a site about Bayes, use an actual Bayesian estimator for ranking/location conditional on votes. The current article locations provide a prior for where articles should be, and the existing votes provide priors for each user. The likelihood functions for how users vote conditional on the quality/topicality/location of a post could be estimated from ordered voting/location history, e.g. P(upvote_at_t | post_belongs_in_main, post_in_main_at_t) != P(upvote_at_t | post_belongs_in_main, post_in_discussion_at_t). P(upvote | posts'_current_score) would be useful for adjusting for priming and follower effects. I don't know how much temporal information about votes/locations is retained by LW. If there's no temporal information stored, at least there would be P(upvote | post_in_main), P(downvote | post_in_discussion), P(no_vote | post_in_X), etc. which is probably still better than purely karma based estimates.

Weight votes based on who voted.

I don't think that would work, because the reason that easy content rises faster is not because the people voting are unable to judge quality.

The upvote grading system is pass / inherently favors content which is just barely good enough to earn the upvote, and is otherwise processed as easily, quickly, and uncontroversially as possible.

Under my model of why easy content rises, Eliezer_Yudkowsky-votes would be just as susceptible to the effect as any newbie LW user's votes...that is, unless high profile users exerted a conscious effort to actively resist upvoting content which is good yet not substantial.

What's worse, you could become a high karma user simply by posting "easy content". That's what happens on Reddit.

On Lesswrong, the readers have a distaste for mindless content, so it doesn't proliferate, but all this means is that the "passing" threshold is higher. So you might (just as an example) still end up with content which echoes things that everyone already agrees with - that's not obviously unsubstantial in a way that would trigger down-votes but it is still not particularly valuable while still being easily... (read more)

Maybe people should be able to give two kinds of karma. One is for "pretty good" and the other (perhaps with a limit to how many can be given in a month) would be for "really excellent".
Some link aggregates have "reaction buttons" - OMG, Epic, LOL, Fail, WTF, and stuff like that. I think it would be cool if someone would make a forum with separate feedback for.. * I personally benefited or learned from this ... TIL * I denotatively think this is factually correct / incorrect ... green checkmark vs. red X * connotative yay / boo... :) vs :( * visibility vote - as in, I want this to be more / less visible to others... open eye vs. closed eye Under this system, the visibility vote would play the function of the primary up-down vote and the "TIL" would be a separate marker of quality, which people could sort by if they wanted. The purpose of the other two feedback forms are primarily to provide an outlet for signalling agree/disagree without adversely influencing visibility - you might want to make a distasteful opinion, common misconception, or well formed argument for an opposing viewpoint visible without creating the impression that it's supported and correct. You might want to make applause or jokes less visible without signaling displeasure or disagreement. Realistically though, a lot of users have talked about deep changes to LW's website functioning in the past, and it doesn't seem like anyone's up for actually doing it. It is kind of a hard job to do.
Alternately, distinguish upvotes ("This is a good contribution") from promotion votes ("This is a significant contribution on an LW core topic").
The UI starts to get really unwieldy if you do that.
At some point, shouldn't content like the latter be identified as either applause lights or guessing the teacher's password? And, theoretically, be documented better in the wiki than the original posts? To me it seems like migrating excellent content to the wiki would be a good way to prevent follow-up articles unless they address a specific portion of the wiki, in which case it can just be edited in with discussion. I haven't spent any time on the wiki, though, which suggests that either I am doing it wrong or that the wiki is not as high-quality as the posts yet. If I imagine a perfect rating oracle it would give ratings that ended up maximizing global utility. If it only had the existing karma to work with, it would have to balance karma as an incentive to readers and an incentive to authors so that the right posts would appear and be read by the right people to encourage further posts that increased global utility. It could do that with the existing integral karma ratings, but at the very least it seems like separate ratings for authors and content would be appropriate to direct readers to the best posts and also give authors incentive to write the best new posts. This suggests both separate karma awards for content and authorship as well as karmafied tags, for lack of a better word, that direct authors in the direction of their strengths and readers in the direction of their need. For example, a post might be karma-tagged "reader!new-rationalist 20", "author!new-rationalist 5" and "author!bayesian-statistics 50" for a good beginning article for aspiring rationalists written by an author who really should focus on more detailed statistics, given their skill in the subject as evidenced by the post.
Maybe, but the idea is to make the best content the most visible. If applause gains higher visibility than content, the system has failed even if the users are able to identify which is which after having seen them both. I think people use the forums more instead of the wiki because the forums are more social and interactive. (I see this is a real benefit of forums over wikis, not as a bug that people have)

One problem I have with this proposal is that Discussion isn't for low-quality content (this seems to be a common misconception), it's for a different type of content. Making Discussion into the land of less popular posts seems like it eliminates potential utility.

When you go to the Discussion page, there is a permanent disclaimer on top: "This part of the site is for the discussion of topics not yet ready or not suitable for normal top-level posts" (emphasis mine). That would indicate that at least originally Discussion was meant for lower-quality content.
This proposal would affect the motivation for voting. Instead of votes meaning something like "I do/don't want to see posts like this", they would acquire a shading of "I do/don't want to see this post in Main".
A modification to take care of this: when a post passes +20 karma, the author opens a poll asking whether it should go to Main, and the vote of the community decides.
I think this proposal deserves serious consideration. I would be surprised if it were adopted, but it deserves serious consideration.
This comment in place of an impossible second upvote.

I am not sure exactly what the problem is, but something must be wrong if even Eliezer seems reluctant to post here anymore (preferring Facebook of all things instead).

Ha ha!

You know, I'd previously assumed it was just him being busy, but now I wonder about those Facebook posts. I assume he doesn't put near as much energy into them as Yvain puts into Slate Star Codex, but still.

It looks fairly obvious that they're much more stream-of-consciousness than his LW content. The difference between RobbBB's recent writeup and the related comment thread on the OPFAI facebook group is marked. I'd take his statement that he finds it an extremely easy medium to get content written in (compared to LW, or essentially any formal writing) at face value.
It seems that the posts on the main blog have become so infrequent that it almost seems as if the site is stagnant, which is definitely not the case. I think having more posts on the main blog would be great for the public-relations side of things.
Most of Eliezer's mini facebook posts are pretty low quality compared to what he used to put on main

I feel like, realistically, people probably just subconsciously rate a given main article along each of the five dimensions you listed (and possibly other ones), and then mash all the ratings together to give a single rating along a general "good/bad" spectrum (which then falls above or below their own main "cutoff"). So a formal set of criteria for main would be nice I guess, but it seems like it would be unlikely to capture exactly what people people mean by "main-ness" anyway - in most cases it's just an intuitive judgement call. I would add, though, that one of the biggest things I look for in a main post (and I don't think I'm atypical here) is insightfulness - presenting new thoughts or ways of looking at things. Which is sooort of captured by factor 2 that you listed, but not entirely.

I rather like this explanation.

We should consider abolishing the distinction altogether.

Main/Meetups instead of Main/Discussion, to merge a couple proposals?

My impression has been that Main is for things that are in the same vein as Eliezer's sequences, while Discussion is more forum-equivalent, for whatever stuff comes to mind or seems interesting today - if you think of LW as a newspaper, Main means you're trying to be a columnist, Discussions are more letters to the editor. Another way of thinking of it is that if someone ever posted an attack piece about LW, I would imagine "That was just some guy posting in Discussion" to be a valid defence, while anything posted in Main would be fair game. (Main postings are moderated, correct?)

Personally, I don't even look at Main.

Main is moderated in the sense that many posts are demoted to discussion, but posting to it does not require explicit permission of the admins. There's also a third section, "promoted," which is a subset of main chosen by admins.

A relatively new user here, my problem is that I have a threshold to posting even in forums that don't have any kind of minimum requirements, like some subreddits on reddit. On LW this threshold is obviously higher and the only place I feel comfortable posting here are those open threads. According to this page there are about 2500 articles on LW and I haven't even read everything in the sequences, I just don't feel there is anything I can say that hasn't been said already.

I don't think this is necessarily a bad thing though, if you're gonna loosen those requirements the quality is probably going to drop because the kind of posts that are "quality" here aren't the first thing in people's minds when they are free to decide what they wanna post.

If you'd like to post about a topic but you aren't sure whether it's been covered, you could ask about it in an Open Thread.

I will try it when the time comes, thanks!
If it's just general philosophical or other kinds of abstract arguments, there probably isn't. If it's about new research developments or new ways of presenting existing content, it very well could be.
2KnaveOfAllTrades10y might also be able to give you immediate feedback from Sequences Black Belts on whether something has already been done to death. Often something which is technically covered by an old post as a special case is still worth looking at specifically or from a different perspective or framework.

We want main to be used for things that are on topic, i.e. things that refine the art of human rationality. I think the main reasons we have few posts on main are 1.) human rationality isn't a very sexy topic, and 2.) actually improving rationality is very hard.

It seems to me that the root problem is not that there are too few posts on main but rather that we aren't collectively making much new progress on the core mission of LW.

I think there a lot of progress made by CFAR on that mission. It just not made by writing lots of articles.
So you think LW should get credit for progress not made on LW and not posted on LW?
I don't think "we" refers to Lesswrong. We are a group of people. I think it's good when everyone acts in a way that produces the most utility without thinking much about where the credit goes.

I think that Main posts should be directly related to rationality, and that posts not directly related to rationality shouldn't go in Main unless you have a really good reason to put them there, especially if you intend to write an extended series of posts. Discussion is for interesting content that isn't directly related to rationality.

That said, it's clear that there's some controversy regarding this issue. Regardless of what the Main/Discussion distinction is or what it becomes, I think we could all benefit from having clearer standards in this respect.

ChrisHallquist, I really enjoyed your posts on Taubes but this was because I'm extremely interested in nutrition. I don't think they strongly enough relate to the core of LW to have several of them in main.

James: I'm beyond the point of looking for opinions on specific posts. Do you mean to imply strong relationship to the core of LW is a necessary condition for Main? How strong is strong? Are the other bullet points in the wiki article also necessary conditions? Those are the kinds of questions I'm hoping I will get answers to in this thread.
I'd view them more like a clustered-condition akin to this sort of thing, where you need some varying strengths of some of the content conditions to activate (not necessarily saying it as simple as a single threshold that they all contribute to, mind you, nor am I saying that those are the only relevant properties.). Frankly, I don't think these nutrition posts score that highly on any of those counts, though. I mean, it obviously doesn't discuss core LW topics, and you've said as much. Going through the others: * The material in your post seems especially important or useful. Well, it's not useless or unimportant - Taubes is a moderately big name in the nutrition discussions I've seen around, and with the paleo folk I've seen, and there probably are plenty of people giving him more weight than they should, but it's not a especially important thing around here is it? * You put a lot of thought or effort into your post. (Citing studies, making diagrams, and agonizing over wording are good indicators of this.) It's good writing, but it doesn't seem particularly high-rigour. I'm aware this might change with future posts, but you can't blame people for not considering unposted articles. If this is the case, maybe a reason to combine posts in the future. * Your post is long or deals with difficult concepts. (If a post is in Main, readers know that it may take some effort to understand.) I don't see this, frankly, and I doubt you do either. * You've searched the Less Wrong archives, and you're pretty sure that you're saying something new and non-obvious. This might be closest. I haven't done the requisite searching, but I doubt this has been gone over before, and while I haven't read Taubes, enough people seen to like him that it may well be rather non-obvious. However, it's not mind-blowing, and with nothing on the other criteria... let's just say that if I were given this list and had to choose based solely on that, this would be solidly in Discussion. Now
This seems very plausible on a descriptive level, but if it's right it's a bad thing for the LessWrong community. Human concepts natural drift towards being defined in a clustery way, but when you're trying to run a community clear rules are extremely valuable.

I think a relevant criterion should be how likely people will still be to want to read and comment the post in five years. (ISTM it's more likely for a new comment to an old thread to be missed by everybody in Discussion than in Main, as the former's Recent Comments page is more diluted.)

Compare with

I was reminded of this recently by Eliezer's Less Wrong Progress Report. He mentioned how surprised he was that so many people were posting so much stuff on Less Wrong, when very few people had ever taken advantage of Overcoming Bias' policy of accepting contributions if you emailed them to a moderator and the moderator approved. Apparently all us folk brimming with ideas for posts didn't want to deal with the aggravation.

Okay, in my case at least it was a bit more than that. There's a

... (read more)

So, I think the bigger problem is not the main/discussion split. It's that, when coming to the site, one doesn't even understand that this is a forum!

I don't know if this is by design, but I was a LessWrong reader for more than a year (!) before I ventured over to the discussion section, and this was largely because I didn't see the link. I realised people were posting things because of the changing articles on the home page, but the same posts kept appearing so I figured there wasn't that much material.

It was only after meeting a few fellow LW'ers at a me... (read more)

I suspect you are atypical, but if not, this is a pretty serious issue.
Hard to say if I'm atypical (Typical Mind Fallacy and all). But I doub tit. Again, I wasn't really looking. But the fact that this is a thriving forum certainly didn't jump out at me form random visits to the site, including the main page. Is there any reason LW doesn't just have a Reddit/HN style homepage?

Eliezer's rule is the real rule: the division is just about quality, as measured by voting. You might ask how closely voting follows the bullet points, but the point is just to predict the votes. Your recent posts don't belong in main because people don't like them. That's all there is to it.

Well, there is also the question of many posts you should have divided it into, but I don't think people really care that much.

If this is true, then we definitely should adopt shminux's proposal. One thing I'm wondering about is how much people didn't like them because I was saying controversial things while also saying, "I will give more justification for this later," which would have been avoided by not splitting the posts.
Did people complain about this? I'm sure you paid more attention than I did, but I only remember this complaint on part 1, the part that was better received. Moreover, I only remember you saying this in part 1. I think the more popular complaint about parts 2 and 3 is that they were not central to Taubes and not interesting topics. But those complaints might have been avoided by having a single post.
I didn't downvote them, but think they would have been a better fit for discussion than for main. While a summary of research or case studies on nutrition would have made a good post for main, I don't think a criticism of a popular book on nutrition (that I haven't read nor even heard much about here) is as generally interesting and useful.
Well, to be frank the reason I don't like the posts in question is that their main argument is the same kind of intellectually dishonest strawman you accuse Taubes of making.
But it is a difference whether an automatism makes this decision based on votes or whether a human pulls the trigger. The current mechanics forces you to consider the implied rules to 'predict the votes'. And the implied rules seem a lot like what Yvain wrote. Maybe not as a hard border but as a more is better with a subjectively perceived threshold. Another difference is that being moved away from Main when you were so bold to post there gives strong negative feedback. On the other hand I haven't seen any Discussion post being promoted to Main since I am here (Sepember). That doesn't mean that I disagree with the rules. I like the quality it ensures in Main. I wonder how many subscribe to Main and not to Discussion. Maybe a question for a survey.
The second most recent post on main, Cognito Mentoring started on discussion. It was promoted not because of the quality of the argument, but because it is an announcement.
Then it must have stayed relatively shortly in Discussion because I didn't see it. But actually that being an announcement means it doesn't really count.
A bit over 2 days from the first comment to the date on the post (which is the date of the move). This should help calibrate your ability to detect these moves. Also, it gives an easy way to detect moves: if the comment is older than the post. Added: also, the meditation post appears to have spent 2 days in discussion. The google cache currently shows the original date; and there is a comment from that time.
Handy. I didn't notice that. I have to assume that I overlooked some moves. I will probably not track that in the future. Tap out.
shrug I subscribe to Main and not Discussion, but that's in part because I only subscribe to things that don't generate more content than I can read or a large proportion of content that I'm not interested in. So Main posts come to me automatically, and then I usually check Discussion manually and just pick out the more interesting-sounding threads
It's not about quality. As the wiki says, discussion is about: * a link with minimal commentary * a question or brainstorming opportunity for the Less Wrong community In other words, stuff where the main focus is on having a discussion. Whereas main is about core LW topics.

Furthermore, it often seems that the "real" rules are significantly different than what the wiki says.

It's a wiki, so you can improve it if you want ;) I revamped the LW faq significantly a while ago, including adding that portion... I was attempting to codify things as they were, not describe how things should be (although, I like the idea of using the FAQ that way as well; if this thread seems to have come to some kind consensus writing it up in the FAQ seems like a good way to codify it for posterity).

I've edited the wiki to read: I do this without any pretense of representing some consensus from this thread; the consensus if anything seems to be that we should modify how the Main / Discussion distinction works at the level of site software. Unfortunately, that doesn't seem likely to happen any time soon. So instead, the above is my attempt to explain the main purpose currently served by Main, influenced particularly by this comment and also Vaniver's comments. It's a rough patch, but I hope it's at least a slight improvement.

Also, while we're discussing meta-level stuff, I think the absolute #1 best improvement that could be made to LessWrong is for a prompt to come up with a comment box inviting you to explain your reasoning whenever you downvote something. Downvotes without comment are IMO extremely bad for site culture and lead to a lot of the frustration discussed earlier.

This would be useful. In fact, I was just thinking about what the best way is for me to solicit anonymous feedback from people who've downvoted me.
I downvoted the parent because I thought it was a bad idea. The unhelpfulness of this comment illustrates why it is a bad idea.
Ironically, this comment was actually extremely helpful. I certainly don't downvote things that I think are bad ideas, only things that I think violate standards of discourse or are inappropriate for the venue in which they are posted. If downvote is just being used as the "I disagree" or "I don't like this" button, I think LessWrong has far bigger problems to deal with!
Why do you believe that explanatory comments work? Here we have an example of an explanatory comment from Larks that failed to communicate anything to you. On Chris's recent series, we have hundreds of explanatory comments that failed to communicate anything to him.
Did you even read my reply? It literally starts with "this comment was actually extremely helpful."
Have you ever seen a productive use of the question "Did you even read my reply?"? Have you ever seen someone believe the answer "yes" to that question?
Yes and yes.
Would you show me?
(Oh, also the reason your comment is surface-level unhelpful is because "I thought it was a bad idea" isn't a real explanation of your decision process, not because explaining votes is inherently bad-- but of course you knew that already.)
My practice is to almost never downvote with comment. If I think something is wrong in an interesting way, then I up-vote it to increase its visibility (because its interesting) and reply with a comment. If I think something is promising but in need of improvement I do the same. If I downvote something, its usually not just because I disagree with it, but because I think its harmful in someway and I want to reduce its visibility. Commenting would tend to promote its visibility and be counterproductive towards that end. If someone is honestly confused as to why they have been down voted, they can explicitly ask and that will prompt me or others to comment.
Downvotes with explanation are way worse than downvotes without explanation. for every post downvoted to -9 I do NOT want to read 9 comments saying "Downvoted for not understanding what the purpose of downvotes is" I don't care if you get frustrated or annoyed at downvotes. That's the WHOLE POINT. As you make comments and continue to be voted up and down you learn the kind of things the community approves and disapproves of, or you get out. Downvoting isn't designed to optimize for being welcoming. It's a moderation strategy designed to improve conversation for the existing participants.
Actually, the more I think about this idea the more I dislike it. It's much easier to learn what is and isn't acceptable behavior if people just tell you rather than you having to suss it out yourself. Further, in many cases the reasoning for downvotes is far from clear, even for established users. I have >4000 karma, I've written multiple top-level posts, I'm frequently on the top 30-day karma list, I know many community members in real life, and I've donated to and volunteered for both MIRI and CFAR-- and even I am not always sure why some posts get downvoted. Imagine how much worse it must be for new users! I've often heard people complain that LessWrong isn't welcoming enough, that it's intimidating, that it's hard to participate in discussions here, etc. In some cases, this simply means that LessWrong's level of rigor is too high for the person making the complaint, and that's totally fine. But I also know a lot of intelligent, level-headed people who are interested in rationality and other core LessWrong concepts, but don't post here because they find the site's attitude towards newcomers off-putting. There are various efforts being put in place to help address this-- for instance, many of Eliezer's original sequences are being compiled into an edited ebook format that should make it much easier for people to access the "core" of LessWrong. But I think that changing either the actual architecture of LessWrong or the culture so that we give more of an explanation for downvotes would be a great help for both existing users and newcomers. Lastly, it's certainly the case that well-kept gardens die by pacifism, and we should be careful about lowering the standards on LessWrong. I think the current standards on LessWrong are part of what makes the site valuable, and I don't want to lower them-- but making them more transparent would IMO certainly help.
I'm primarily concerned with downvotes that nobody explains at all. As Alsadius pointed out, if someone else has already explained their reasoning well, you can simply downvote the main comment and upvote their reply. But there are many posts that I see being downvoted for unclear reasons with no replies at all. I've also noticed that when I call this out by making a comment that says "why is this being downvoted," the voting trend often reverses. This strikes me as a problem.
A year or two ago I tried replying more often with "downvoted because...". This did not appear to have any effect beyond some karma penalty for me when my reasons for downvoting were the least bit controversial.
I've occasionally written well-received downvote explanations, but that only seems easy to do when the reason the parent's being downvoted is or should be obvious to all and sundry. I gather from this that the community likes seeing people scolded who make dumb points, but doesn't like downvote explanations in general. (I'll admit a fairly small sample size, though -- typically I don't downvote posts I respond to, because of the risk of sparking a retaliatory cycle.)
I don't do this all the time, but I'd say about half my downvotes are accompanied by a comment arguing with the post, though not always mine if someone else has made the point well. If my complaint is non-obvious, I try to do it every time.
This idea has been reinvented and discussed multiple times over the years, and it might be a good one, but Eliezer and other admins are not interested enough to spend any of their limited Trike resources on this.
That's unfortunate, though it might of course still be possible to accomplish this shift on a cultural level or with more minor code changes.
I agree. Or perhaps a dropdown box with categories describing the general reason for upvote or downvote.
Well, Eliezer's attitude appears to be the exact opposite, he even implemented a 5 karma penalty to discourage people from replying to down-voted comments.
Heavily downvoted comments, yes. This proposal is more oriented towards everyday voting situations.

I'd like to go further and get rid of the distinction between main/promoted, main, and discussion (I'd probably keep open threads, though). Just post articles, let them get voted up or down based on whatever "the community" wants to see, and allow a few high-karma individuals to promote things they want to see on "the front page" as our face to newbies or casual readers.

Among other things, that would discourage highly-skilled newcomers from posting anything, and it would encourage people to pursue karma at the expense of quality or clarity (just look at Lumifer's posts).
How would this discourage posting? Seems like it would make it easier for high-quality posts to get promoted (because they're posted with the same bar as discussion has today, and can get selected for promotion directly from there). I should probably have been less specific about who gets to promote things - "site founders" or "a few designated site ambassadors" is mostly what I was thinking, not "anyone who collects enough karma". I'm not sure how one pursues karma at the expense of quality or clarity - don't unclear or low-quality posts get downvoted enough?
If that's the case then I misunderstood you. By posting far too often, with the hope that posts will get upvoted more than they get downvoted.

I think the main/discussion divide is useful, for a good purpose that isn't immediately obvious if you are a long-time user, which is highlighting the best stuff that LessWrong has produced throughout its history for newcomers first. If they are hooked, then they keep reading until they exhaust the content of the lowest quality they find worthwhile. The people who read LessWrong the most are the people most interested in the stuff here, and therefore most willing to put up with lower-quality posts. Main/Discussion is meaningless to them, because they will ... (read more)