LessWrong Boo Vote (Stochastic Downvoting)
Related to: Well-Kept Gardens Die By Pacifism.
I wrote a script for the greasemonkey extension for Firefox, implementing less painful downvoting. It inserts a button “Vote boo” in addition to “Vote up” and “Vote down” for each comment. Pressing this button has 30% chance of resulting in downvoting the comment, which is on average equivalent to taking 0.3 points of rating. If pressing the button once has no effect, don’t press it twice: the action is already performed, resulting in one of the two possible outcomes.
The idea is to lower the level of punishment from downvoting, thus making it easier to downvote average mediocre comments, not just remarkably bad ones. Systematically downvoting mediocre comments should make their expected rating negative, creating an incentive to focus more on making high-quality comments, and punishing systematic mediocrity. At the same time, low penalty for average comments (implemented through stochastic downvoting) allows to still make them freely, which is essential for supporting a discussion. Contributors may see positive rating of good comments as currency for which they can buy a limited number of discussion-supporting average comments.
The “Vote boo” option is not to be taken lightly, one should understand a comment before declaring it mediocre. If you are not sure, don’t vote. If comment is visibly a simple passing remark, or of mediocre quality, press the button.
How about two standards?
For top-level comments, downvote mercilessly, as per Eliezer’s recommendations, so readers with little time can quickly find the most interesting stuff.
For comments-on-comments (all the little threads), be more lenient—downvote only trollish, rude or stupid comments. “Sorting by votes” is less useful in threads anyway. That would make them a better place for chit-chat, questions, etc.
I would base the division on length—long comments should be downvoted mercilessly, as they take up more time and more screen real estate—but this makes sense too.
A broad community norm against long comments is probably a good idea, anyway. The vast majority of long comments are likely going to be either off-topic, a good idea expressed too verbosely, or something with enough length and quality that it probably should have been a full post.
I’m pleased to hear someone say that. I worry that my comments are too short, but I do often put work into brevity.
I have the opposite intuition. I prefer comments that present a full argument, preferably with links and/or references—those are the ones I tend to upvote. Long comments that drift off-topic, I downvote for being off-topic, not for being long.
I think you misunderstand the point. Useless comments should be downvoted more readily the longer they are as that means they are a less efficient use of pixels. Shorter comments don’t need to be downvoted as much even if they’re not the best since they’re not polluting the screen as much. In general, the best comments will take up some space but I take it there is always some length at which even a helpful comment deserves to be downvoted.
My intuition is similar, and for similar reasons.
Agreed, I thought about adding that as a second alternative, but skipped it to prevent my comment from getting too long ;-)
Length isn’t as much of a problem when the discussion has turned into a back-and forth between two commenters, or in a reply to a question, but still, overall, we should discourage a long, rambling style.
I like this as a general improvement, but it doesn’t seem to solve the problem. The kind of trolls that write filler comments and nothing else will just shift in the comment-on-comment niche, making it uncomfortable. The point is that ultimately, there needs to be some kind of moderation of people.
I endorse this idea strenuously. Aggressive housekeeping on all comments will suppress participation, cf. Alicorn’s post, but this seems to be an excellent compromise.
How hard is it to modify the point system? For instance, could we track positive and negative votes separately? This would make it easier to find, for example, controversial posts: high positive and negative votes. Those are the posts I’d personally be most interested in reading.
I’m fond of the Amazon rating system. Seeing a simple spread informs me much better (and just as easily) of potential value than a single, all inclusive number.
Data point: I was going to make a different comment, but have been made paranoid and inclined to hoard my existing karma by the new, more liberal downvoting recommendations. I’m never going to find out whether my original comment would have improved anyone else’s reading experience or not, because I didn’t care about it enough to take a newly increased risk as the price for saying it.
I don’t want to trade in karma I earn for great comments so I can make mediocre comments. I want my karma to go up, preferably fast. I’m not sure if this is a good goal to have—maybe karma just feels like approval and I like it because I wasn’t hugged enough as a child or something—but nevertheless, I have that motivation.
I advise against this strategy. The quality of your comments is generally positive, and that will be reflected in your karma, so just post when you have something to say and it will go up overall, though it’s pretty unpredictable what will be voted up or down.
And yet… the strategy your instincts are driving you to implement isn’t making your karma go up fast. At least, it would go up faster if you lowered your standard. What your extra caution is achieving in practice is sacrificing karma gain in favour of conveying a reputation as a considered, insightful poster with a diplomatic style.
If you actually want to increase karma faster, post a whole heap more and only inhibit posts that are rude or particularly stupid.
That said, it is possible to maintain high karma and come across as an entirely unhug-worthy pain in the ass a lot of the time. Data Point: The most downvoted person on this site remained in the top posters list until a couple of weeks ago when I spent ten minutes browsing his posts making sure I that I downvoted everything stupid. By all means, I encourage you to continue to err on the side of not being a pain in the ass.
I’m still making comments. And I value some strategies over others to get karma—I don’t want it to go up for sheer volume and dumb luck. I’d rather make comments that come to mind as good things to say, not just type the unfiltered contents of my brain. (The unfiltered contents of my brain contain a lot more fiction references, inside jokes that are known only to my close personal friends, and musings about muffins and tofu and broccoli than anyone wants to read.) And I’d rather the good ones get voted up and the bad ones get informative replies—not freely dispensed downvotes—so I can adjust my behavior with more information than a brute negative number.
Also, as a counterexample, I took a karma hit for a number of comments on another thread—I remarked on it and they’ve all been compensated for, but if I hadn’t said anything I imagine they’d have sat there indefinitely, because apparently the new “downvote freely” ethos hasn’t made me paranoid enough.
For comparison, I prefer to comment the same way I would if nobody was voting on them. If the people here like my comments, then they’ll float to the top. If my comments routinely got downvoted to oblivion, I’d probably post elsewhere instead.
In other words, I’m a big fan of the marketplace of ideas, but I don’t have one in my own head.
Insofar as voting is collective feedback from this community, it carries at least some informational value that one should value to the extent that one considers the average LW user to have intelligent, insightful opinions.
It does. Perhaps the most significant value I get is in calibrating my expression. People here, who for most part are a relative inferential distance from myself and relatively high in comprehension ability. When people here don’t understand what I’m saying then chances are I’m leaving too many inferential steps implicit, expressing myself inelegantly or conveying concepts too quickly and I need to adjust my communication.
In context, if these grammatical mistakes were intentional I tip my hat to you, sir. Well played.
Is that bad?
Considering that some posts are getting hundreds of comments, not that many people have the time to read them all (especially if you have to search a bit to find what you have and haven’t read), it may be better for everyone to have fewer comments, but of higher quality.
Or, to put it another way, considering that you’re writing once to be read dozens of times, it’s nice to your readers to take a bit of effort to polish up your prose, it costs a few seconds to you but can save a few seconds to a lot of people. This may feel unusual if we are used to situations like conversation (or online chat) where the listener/talker ratio isn’t as skewed.
The real risk is when certain forms of comment (approval, disapproval) are discouraged, because the community’s standards of “quality” are skewed.
Agreed. I strongly feel that comments of a few words expressing thanks, agreement, apology, sympathy, approval, acknowledgement, etc. should simply hover at zero. Such remarks are part of the native architecture by which we communicate, and I think we lose something if we discourage them.
I agree with the descriptive content of what you wrote, but not the normative content. I agree that we do lose something if we discourage these sorts of comments. However, short comments that don’t add anything to the discussion (like the ones you mention) do add a significant amount to what gets displayed on the screen. If someone is reading this with a screen reader, a text browser, an iphone, or even just a small laptop or old, low-res display, then they will have to wade through “MBlume 22 April 2009 07:55:59AM* 8 points [-] Thanks—I agree. Vote up | Vote down | Permalink | Parent | Report | Reply”, for no good reason. Much better to discourage this sort of noise that adds nothing to the pursuit of rationality as such.
Might be bad—in principle, it would be nice if we could sift through many comments as a community, so as to increase the number of good ones that can float to the top. If everyone who reads a bad comment votes it down, not that many will be inconvenienced by additional mediocre comments, and allowing folks to attempt comments they think might be good (and might not be) would plausibly increase the absolute quantity of good comments. (This depends on our ability to sort good comments to the top and mediocre ones to not-much-read locations, though.)
A second reason it might be bad is that commenting, and engaging with LW content more generally, increases the chances that the commenter will learn from it and do something with it. But, again, the costs may be prohibitive if such comments stay mixed with the best of LW.
It’s not a matter of polished prose. I’m not paranoid because I think I might make a grammatical error or abuse semicolons. It’s a matter of what ideas I commit to words and send out to the community. My thought process when I was debating whether to scrap my original comment was anxious and convoluted: “There aren’t any other comments yet, do I really want this to set the tone for the entire discussion? This idea relies on the kind of personal anecdote that’s gotten me poor results before. But it’s still relevant, and I’ve been mistaken in the past about whether a comment would be well received...” At which point I decided that even if the original comment wasn’t good enough, my internal agonizing probably was.
Well, karma is a tangible (if imperfect) measure of success in this community, and rationalists should win… right?
I think this perception is a problem for the community, and I don’t think it’s workable to tell people to not feel like this. Two possible ways around it:
Make comment scores affect karma only if it goes outside some range, e.g., perhaps you’d lose a point of karma for a −2 post, two points for −3 post, gain a point for +2 post, two points for a +3, &c. This would likely be a pain to implement.
Make karma dropping normal and expected, but not tied to participation, e.g., take the square root of everyone’s karma on a weekly basis. This has obvious downsides for karma cutoffs to do certain things (like post articles).
I like this conceptually, but pragmatically I do not see its advantage. I think one way to do something similar is to keep karma split into positive and negative karma. I am currently around 75 karma, but my hunches tell me that is probably +125 −50. Someone sitting at +125 −50 is different than +250 −175 or +80 −5.
Having bad karma get its own bucket is less demeaning than having it change your entire “score”. It is still bad but you can still get a feel good from the positive scores.
(Edit) Oh, Jordan said the same thing. Please ignore this comment.
See the Preferences button next to your karma counter? Use it! Raise your hiding threshold instead of devising convoluted ways of mass-semi-down-voting-whoop-whoop mediocre comments bang!
This doesn’t make sense to me. If a comment supports the discussion, it adds value to the site and should not be discouraged. Downvotes should be used to discourage comments that disrupt the discussion.
One point is already a small increment. This is as silly as choosing to cast 30% of one vote for a candidate in an election.
I don’t see how this analogy applies. The problem that is being solved is keeping proper balance between good comments and average comments. Average comments must weight down, as otherwise there isn’t enough feedback, but they mustn’t weight so much as to hurt normal discussion. This can’t be solved with extremes within the current system.
The analogy breaks because, generally speaking, voting for a candidate in an election is mutually exclusive. You only get to pick one. It does not matter what the weight of the pick is because everyone has to pick with the same weight.
Here, you can choose as many comments as you want and choose different weights for each pick.
Those of you excited about this: aside from the presumed difficulty of implementing it, would it be even better if there were an option to actually vote −0.3 on a post, instead of voting −1 with 30% probability? And would it be even more of an improvement if you could choose to vote anywhere in the [-1, 1] range, so that you could mark something −0.7 or +0.25?
Those suggestions probably seem like an exaggeration, but I really do think we’re all getting too worked up over the minutia of the karma system. This isn’t a game. These numbers aren’t our high scores. It feels like there’s too much temptation to regard them that way, and further complexity to the system will only increase that.
Agreed, but:
We must admit that to a great extent, it is. We are all attempting to make ourselves appear more useful to the community, and karma is the only quantitative way to tell if we’re making progress. Like so many things, it feels like it trivializes but it is there for a purpose.
That gets to the heart of why I don’t think the karma system is worth too much emphasis. Shouldn’t we instead be attempting to make ourselves more useful to the community?
That’s true. I do think we’re better off with it than we would be without it, but it shouldn’t get attention disproportionate to its purpose. It’s a means to an end, nothing more.
That’s the thing. Controlling things with ‘shoulds’ is unstable without the presence of real consequences, social or otherwise. Anonymous internet forums do not have these real consequences naturually, which is what gives Karma a purpose. It is a way to allow social control and influence with the minimum of overhead and perceived oppression.
Right. The whole point is that what karma really controls for is appearing useful to the community, not being useful to the community.
I agree that it has a purpose, and that we’re better off with it. I don’t think it’s sufficient on its own, and we shouldn’t fool ourselves into thinking that obsessing over it is the same as focusing on improving the community. At best, it improves only a small aspect of the community; at worst, the subgoal “think about karma and get points” takes over at the expense of all else.
Why do we feel that a downvote is such a big deal? A downvote doesn’t mean “we think your an idiot, go jump off a bridge or at least don’t ever post here again”. It means that this comment isn’t up to our standards, and you should do better next time.
It’s a big deal if it’s the first or only vote on a comment, because it puts the comment into the negative. A negative score by itself presents the message, “the community considers your contribution a net negative”.
I suspect, though, that if the same score were displayed as “+0/-1”, it would communicate a VERY different message, i.e., “One person didn’t like this, and nobody else cared one way or another.”
Those two messages are likely to produce different effects in most people’s brains.
A downvote doesn’t mean ‘we’ anything. It means “I disaprove of this comment”.
hrm. yes. you’re right. a score means “we” though.
If I understand correctly, the distinction between not voting, upvoting, and downvoting is too broad. A -.3 vote would give a fourth option that allows for “Didn’t suck, but not great.”
Why a 0.3 chance? Is that totally arbitrary? Also, it seems like a “boo” button would quickly become a means for people to indulge in inappropriate down-voting and feel insulated from responsibility for the outcome. It would also be a tempting false compromise between actually down-voting and doing nothing. Usually, one or the other is the right choice.
Wow. When I speculated creating a stochastic vote script I more had in mind a was one that made each vote of mine worth an expected value of three or four.
There is a point where my utility is improved by allowing other people’s insights, opinions and judgements to make the deciding impact on whether a comment is sent to truncation hell or to the pinaccle of (aparent) public acclaim. It seems likely, however, that my own preferences are more likely to be served were my votes to count as more than one rather than less.
It just so happens that the effort of developing a system that can magnify my own voting powers without being obvious would be greater than the benefit to me. As such, I haven’t coded it. Three votes for Micro-Economics!
boo