LESSWRONG
LW

Comment Permalink

10y20

I think people acquire a belief that a post or comment of a certain felt quality deserves a rough number of upvotes or downvotes, so they don't add or subtract karma when the post or comment hits that level.

Sounds familiar and could indeed explain why some posts do not continue to accumulate votes after some time.

Let's check:

I think a post deserves a certain number of votes/karma and up/downvote accordingly[pollid:939]

Capla

10y90

I generally don't care what level a post is at if I'm going to upvote it, but when I see something has a negative core that I think is unfair, I'll bump it up by one.

See in context

29 Guidelines for Upvoting and Downvoting?

by Sable

6th May 2015

1 min read

29

I've only recently joined the LessWrong community, and I've been having a blast reading through posts and making the occasional comment. So far, I've received a few karma points, and I’m pretty sure I’m more proud of them than of all the work I did in high school put together.

My question is simple, and aimed a little more towards the veterans of LessWrong:

What are the guidelines for upvoting and downvoting? What makes a comment good, and what makes one bad? Is there somewhere I can go to find this out (I've looked, but there doesn't seem to be a guide on LessWrong already up. On the other hand, I lose my glasses while wearing them, so…)

Additionally, why do I sometimes see discussion posts with many comments but few upvotes, and others with many upvotes but few comments? If a post is worth commenting on, isn't it worth upvoting? I feel as though my map is missing a few pages here.

Not only would having a clear discussion of this help me review the comments of others better, it would also help me understand what I’m being reinforced for on each of my comments, so I can alter my behaviors accordingly.

I want to help keep this a well-kept garden, but I’m struggling to figure out how to trim the hedges.

Personal Blog

29

New Comment

63 comments, sorted by

top scoring

Click to highlight new comments since: Today at 8:52 PM

[-]Unnamed

10y270

An upvote communicates to other readers "this comment is worth your attention."

If a comment is more highly upvoted, people are more likely to read it rather than skip over it, more likely to read it closely rather than skim it, more likely to follow links that it contains, and more likely to spend some time thinking about its arguments rather than just moving on.

Downvotes sort of do the opposite, but it's not perfectly symmetrical because scores below zero pack an extra punch.

[-]itaibn0

10y60

Downvotes sort of do the opposite, but it's not perfectly symmetrical because scores below zero pack an extra punch.

The standard guideline is to upvote if you want more of that kind of comment, and downvote if you want less. The asymmetry between upvotes and downvotes comes the fact Less Wrongers on a whole want more content on Less Wrong rather than less. Negative scores pack a punch because they mean your comment would be better off not existing.

Well really, I think it's mostly that people just have a pre-existing idea of the connotation of negative numbers, but I gave this retroactive justification to show that I think the result is surprisingly internally consistent.

[-]John_Maxwell

10y60

The asymmetry between upvotes and downvotes comes the fact Less Wrongers on a whole want more content on Less Wrong rather than less.

In principle one could tweak the interest rate on the Less Wrong content economy by eliminating the ability to downvote for users who have less than X karma. Personally, I think we are currently too harsh in our moderation and backing off a bit (requiring 1 karma to downvote) would be a worthwhile experiment. Less Wrong is already going to rip your ideas apart plenty without the added injury of downvotes. The bigger problem is people being Negative Nancys about unimportant things like whether your thing would be more appropriate for discussion or open thread or what have you. Enforcing rules has a social cost; frivolous rules mean you're paying the cost for little gain. I also dislike seeing people downvote posts past -4 or so for no real reason (god forbid anyone should attempt humor on April Fool's day).

Another idea I had is trying to create a norm of people sending appreciative private messages for posts they like. I think posters would be extremely responsive to this sort of social reward (much more than mere upvotes, which invite score comparisons etc.) and since they're private messages, they wouldn't clog up the site with content-free praise comments. This could be done informally, or it could be institutionalized if there was a popup that would randomly sometimes appear when you upvoted something inviting you to send an appreciative PM. I actually think this could be pretty fabulous for Less Wrong if it was done right. I don't think people realize on a System 1 level how fantastically valuable a thoroughly researched post on important ideas is, but right now writing that sort of post is a thankless job. (Or: "When making a correction or complaint as a top-level comment, choose one positive thing about the post, if any, and point it out first." Or maybe we can just embrace the idea that harsh criticism is normal and expected?)

[-]Gunnar_Zarncke

10y50

Downvotes sort of do the opposite, but it's not perfectly symmetrical because scores below zero pack an extra punch.

Indeed. I have 'reversed' quite a few '-1' because of this.

[-]gjm

10y130

I have upvoted more comments that are on -1 than I have downvoted ones on +1 (even taking into account that I upvote more than I downvote), but for a different reason: what looks to me like an asymmetry between unreasonable upvotes and unreasonable downvotes.

Most of the time when I do this, it is because I see something obviously reasonable sitting on -1, think "what the hell?", and conclude it must be an ideological downvote where someone is voting someone else down for being visibly on the wrong team. And I don't see obvious signs of ideological upvoting.

However, this may simply indicate that ideological downvoting is easier to spot, because downvoting is generally rarer (so if an OK-but-unremarkable comment is at -1 then that's more evidence of malfeasance than if an OK-but-unremarkable comment is at +1). Which, in turn, is probably because of that psychological difference between positive and negative, and/or the idea that "negative scores mean your comment would be better off not existing". So maybe, on sufficiently careful analysis, this comes down to the same effect after all :-).

[-]lululu

10y00

Yeah, I've been downvoted to a negative number in Quora with a nice, detailed, science based article about why someone's "recently recovered early-childhood memory" were probably not reliable enough for her to publicly accuse someone of molesting children and without first talking to a counselor and preferably a councilor who understood memory. It was all very reasonable and with ample evidence to support every statement.

I got down-voted to negative by a guy who said she should try a past-life regression and literally used the phrase "sort of a quantum parallel worlds reincarnation metaphysics with possibility of memory across various worlds and/or lives". (Actually, if you guys could upvote me on that, I'm genuinely worried that other people with similar recovered images will find his advice and be unable to see my comment and that lives and relationships will be destroyed, the thread is https://www.quora.com/I-think-I-have-been-abused-sexually-as-a-child-I-just-have-some-faint-memory-picture-which-came-up-very-recently-in-my-mind-How-can-I-be-sure-I-am-a-20-year-old-female)

[-]TheMajor

10y270

If you want to read more posts like the one you just read, upvote. If you want to read less posts like the one you just read, downvote.

[-]TheAncientGeek

10y100

That's a bad criterion, since it's an open gateway to confirmation bias. Having your biases challenged feels bad.

[-]John_Maxwell

10y90

I've argued that counterfactually we should go a bit easy on unpopular views, even if the reasoning seems incorrect on further examination, since it's hard to predict in advance whether someone will find flaws in your reasoning, but they'll go especially hard on you if you present flawed reasoning for an unpopular position (relative to a popular one), discouraging arguing for unpopular positions.

[-][anonymous]10y10

To be fair, I think a more correct response would be that TheMajor's statement is correct, but implies a criterion which would likely lead to confirmation bias. Applied correctly, a better understanding of one's wants would support TheMajor's conclusion because one would not want to just have one's views confirmed.

[-]TheAncientGeek

10y10

In other words, it works just fine, so long as you are in an ideal world,

[-]TheAncientGeek

10y10

Case in point

[-]Gunnar_Zarncke

10y30

This is an example where I - after some deliberation - decided to downvote. I believe this to be grossly overvoted. I agree that it is a simple and handy rule of thumb. But 22 upvotes? It isn't that insightful. It isn't a that much more correct choice than the approches mentioned in other comments.

Note that I wouldn't have downvoted if this comment would have stood in isolation. I see this post an example of a place where the votes order the comments in relation to how well they answer the question of the OP (which is facilitated by the default sort being 'best' voted). Actually a pattern not mentioned in the other comments.

Lets poll what you think of this:

If a post asks for advice (like in the OP) I vote depending on how well the comment answers the question. [pollid:952]

[-]NancyLebovitz

10y140

I'm not sure it makes sense to upvote all the articles you comment on, though maybe it does if you like the post and you're doing a direct comment to it. Comments to comments are as likely to be a discussion between the commenters as directly abut the post.

I recommend downvoting things which are malicious and/or incoherent. I can see a case for downvoting things which add little information (except jokes-- it seems to be part of the culture to upvote things which are funny), especially if they're taking a lot of words to not add information.

I think people acquire a belief that a post or comment of a certain felt quality deserves a rough number of upvotes or downvotes, so they don't add or subtract karma when the post or comment hits that level. I might just be extrapolating from myself on this, but I think posts and comments, especially comments, tend to hit stable karma levels fairly quickly. However, I wouldn't put off giving/deleting karma until you've got that sort of felt sense of typical karma.

If someone mentions in a comment that they've updated a belief, they get some upvoted. This is something I like a lot about LW culture. It's also a handy thing to tell people if you want to explain something likable about rationalism.

[-]Gunnar_Zarncke

10y20

I think people acquire a belief that a post or comment of a certain felt quality deserves a rough number of upvotes or downvotes, so they don't add or subtract karma when the post or comment hits that level.

Sounds familiar and could indeed explain why some posts do not continue to accumulate votes after some time.

Let's check:

I think a post deserves a certain number of votes/karma and up/downvote accordingly[pollid:939]

[-]Capla

10y90

I generally don't care what level a post is at if I'm going to upvote it, but when I see something has a negative core that I think is unfair, I'll bump it up by one.

[-]John_Maxwell

10y120

The Less Wrong FAQ (linked to from the About page) is meant to address questions like this one. It's still worth revisiting them periodically, though. In order to codify any conclusions we come to for future generations, I suggest making your proposals in the form of polls where the options correspond to changes in FAQ text. (You may wish to include a "I just want to see the results" option in your poll, BTW.)

[-]Stefan_Schubert

10y91

Thanks. The instructions are quite vague, though. On the one hand, it says:

Please do not vote solely based on how much you agree or disagree with someone's conclusions.

On the other:

In some cases it's probably acceptable to vote in order to register agreement or disagreement. For example, you can vote someone's proposal up or down based on whether you think it should be implemented.

In my view, this gives too much leeway to people to vote purely on the basis on disagreement with the conclusions.

[-]Lumifer

10y50

this gives too much leeway to people

There is no enforcement mechanism, so people have all the leeway the want regardless of what the instructions say.

[-]John_Maxwell

10y10

Should we delete the "In some cases..." clause?

[pollid:951]

[-]Elo

10y00

repeat poll in a more public location?

[-]IlyaShpitser

10y120

Karma is both a very noisy and a very biased signal -- it takes too much discipline from too many people for it to be otherwise.

Very negative and very positive karma scores have some signal, otherwise probably safe to ignore.

[-]TheAncientGeek

10y50

I don't think it's an Iimpossible standard to separate "how well did the author argue their point" from "how much .I agree with the point the author argues".

[-]IlyaShpitser

10y20

I am sure it's not, but people in an online community, even this one, will not meet it in large enough numbers. It's hard to be rational.

This is not meant to be judgemental, I am sure I am bad at this too (but I pretty rarely use the karma system).

[-]TheAncientGeek

10y00

The results will fall short of perfection but that isnt a good reason no to try. Fallacy of gray.

[-]IlyaShpitser

10y00

I am not saying you shouldn't try, I am saying given the current state of things, it is probably safe to ignore any karma score that isn't very positive or very negative. It's hard to get signal from other scores.

Let's say a post of mine has -2 total karma. What should I conclude from this? My point is, essentially nothing.

[-]Nornagest

10y100

It's mostly up to you. The standard advice is "upvote things you want to see more of, downvote things you want to see less of", and that's not a bad starting point especially on the upvote side, but there's a little more to it than that.

People respond more strongly to downvotes than upvotes, so most people don't downvote merely low-quality posts even if they'd like to see less of them; that takes something more severe. I tend to downvote stuff if it seems to have been posted in bad faith, or if I find it rude, unreadable, obnoxious, or excessively partisan; I'll also sometimes downvote top-level posts (not comments) if they're poorly written or uninformative, especially in Main.

Downvoting posts because you don't like the people that wrote them is explicitly against policy. Don't do that. On the other hand, you'll probably find yourself downvoting the same people a lot, and that's fine as long as you're basing your judgement on the posts themselves.

Incentives-wise, there's one major caveat, and that's that comments (but not top-level posts) downvoted below -4 can no longer be responded to without paying karma of your own, and neither can any of their children: the so-called troll toll. If you feel a comment that's being downvoted deserves further discussion, you might not want to downvote it below that threshold even if you otherwise would.

I've also found it's a good policy not to downvote responses to your own posts; it helps keep conversations civil and it helps avoid messy and unproductive downvote duels in long threads.

[-]Sabiola

10y20

I upvote responses to my own comments on principle. Unless they're nasty, but that hasn't happened yet on LW.

[-]Stefan_Schubert

10y90

Lots of people seem to downvote a comment or a post simply because they do not agree with its conclusions, as was discussed here. That's wrong, in my opinion. Instead, you should only downvote in case the reasoning is poor, there are personal attacks, and so forth.

[-]Gunnar_Zarncke

10y80

Some polling about voting rules/habits:

Posts and Comments should be voted on depending on quality [pollid:940]

Posts and Comments should be voted on depending on how much you like them [pollid:941]

Posts and Comments should be voted on depending on how much you agree with them factually [pollid:942]

Posts and Comments should be voted on depending on how funny they are [pollid:943]

Posts in Main should be held to very high standards with respect to quality ('scholarly') [pollid:944]

Posts in Discussion should be held to higher standards than comments [pollid:945]

Posts may specify special voting rules to apply in the comments (there are a few precedences for this e.g. the survey) [pollid:946]

Downvotes should only be given for violations of community norms [pollid:947]

Downvotes should be given symmetrically to upvotes [pollid:948]

Posts and comments by trolls should be downvoted [pollid:949]

Ceterum censeo polls are a quick and easy way to quantify opinion.

[-]jbay

10y40

Upvotes and downvotes should be added independent of the post's present score [pollid:950]

[-]Gunnar_Zarncke

10y60

I do not only honor abolute quality but also honor effort with votes. For example newbies gain upvotes easier because I honor the effort for trying to fit in. I upvote references and summaries. And I upvote chores like posting period threads and sometimes meetups (seems like there are exactly 5 people who do that).

And apparently I honor polls with votes what few other people do. I do not know why people do not upvote a poll on which obviously many people participated. Did they expend their click energy with the poll itself?

[-][anonymous]10y60

Sometimes I have trouble deciding how to vote. Many times I upvote as a "thank you" for commenting on something I posted and sometimes I upvote because I agree with the comment.

Also, I upvote when I don't agree, but the effort of writing and explaining their opinion is so well structured.

Downvote only when it's missing the point or disrespectful.

[-][anonymous]10y40

I'd prefer unedited mind dumps not to be downvoted, mainly for the reason my articles tend to be ones. I believe if a forum title says Discussion, we can have more equality of effort. I.e. in normal articles the authors should put a lot of effort in making it easy to read and the reader should not need to put in much effort. But in something called Discussion just dumping loosely related ideas, insight-beginnings, and asking the readers to put in some effort into processing it, fishing out useful sounding bits, and engaging with them, should not be too much IMHO. If you don't feel like doing so, fine, just ignore it altogether, but hitting a downvote sounds like sending everybody the message that they should not either, and sending the author the message to not do this at all. Which is IMHO wrong, these things have positive utility, in the sense of a chance of having some bits that could be developed into something insightful.

[-]John_Maxwell

10y20

It might be worthwhile for you to edit your mind dumps given that the payoff will be multiplicative across your readership. Let's say you spend 20 minutes editing your mind dump and compress it to 50% of its original size. Maybe originally it took 10 minutes to read and now it takes 5. If your post has 100 readers over its lifetime (quite plausible to me given the number of lurkers on Less Wrong, the way people will stop voting once a post reaches its "just" score, and the high number of comments many posts get), then we are 480 minutes ahead of where we were. (Incentivizing people to do things like this seems like a positive side effect of voting.)

[-][anonymous]10y20

Thanks, this actually makes sense.

[-][anonymous]10y40

I downvote whenever I feel a comment was made in bad faith; trolling in other words. This means I rarely downvote. If I disagree with the statement or believe an error was made, downvoting doesn't seem too helpful as the person will likely have no idea why the downvote was received. I'll sometimes even upvote what I think are errors as an interesting error that provides good discussion can be quite valuable. If you never make a mistake, you're not pushing yourself enough.

I've noticed more controversial topics tend to receive more downvotes even adjusting for quality of content.

I've wondered what percentage of the community hands out the majority of the upvotes and downvotes; especially downvotes because of their rarity.

A good chunk of karma votes are just pack following.

[-]Lumifer

10y40

Don't know about guidelines but I tend to downvote for stupidity (IMHO, of course) and rudeness, and upvote for insightfulness and unusual thinking.

[-][anonymous]10y40

Upvoted :)

I'm new here too, and have been curious about this. Like you, I've been hesitant to give down votes. But I think that if I ever down vote, it will only be when (1) someone seems inappropriately confident in what they're saying or (2) a comment was needlessly (and it's always needlessly) mean and condescending to someone else.

Also, I wouldn't down vote a reply to one of my own comments, since I might be biased. I'd down vote only as an objective observer of other conversations.

As for upvotes, I'm a bit more generous with discussion posts than comments. It takes more time to come up with a good post than a good comment. For longer posts, it would be cool to have an option to give the writer between 1-3 karma. I sometimes reread posts from the Rationality ebook on here and am surprised to see they have only 30 or so karma points and wish I could give them more than one thumbs-up.

[-]Dentin

10y40

There are many other categories suitable for downvoting which I've seen over the years. I've seen blatantly incoherent streams of thought, comments containing 'facts' known to be flat out wrong, posts with an extremely high percentage of spelling and grammatical errors, and material I simply don't want to see here (rape and pedophilia related stuff springs to mind.) These are just the categories that spring to mind at the moment.

Only downvoting based on confidence and 'being mean' may cover most cases, but it still leaves a lot of significant bases uncovered.

My upvote/downvote rule has been "Do I want to see more of this type of thing on LW or not?" The three possible responses that I use are yes (+1), no (-1), and 'meh' (0).

[-][anonymous]10y00

Yeah, you're right, those are more good reasons, and I'd probably downvote for them too!

[-]SarahNibs

10y40

Different folks have different reasons, and it's not just a noisy-but-additive result, since some folks use an algorithm like "upvote an article that 'feels like' an article that should end up at +5 if it's currently below +5, otherwise don't, or even downvote" for whatever value of 5 the article makes them feel. A control system in the midst.

[-]Username

10y30

Respond to this anonymously if you want (with the Username/password account): what are some inappropriate reasons you've upvoted or downvoted something on Less Wrong?

[-]Username

10y170

There are some comments which I wouldn't normally upvote or downvote, but if I think they've been upvoted or downvoted disproportionately to their value, I'll try and redress the balance. For example, if someone asks a banal but well-meaning question and it gets downvoted, I'll upvote it.

[-]NancyLebovitz

10y70

I consider this marginal rather than flat-out inappropriate, but I've upvoted posts to avoid paying a karma penalty for replying to them. Someone I've discussed this with withdraws the upvote (click on the boldface thumb to withdraw a karma point) after having commented.

[-]Sable

10y50

I've mostly tried to avoid upvoting so far, and I've completely avoided downvoting.

My model for upvoting right now is:

If I've commented on a post, I should upvote it, because if it was good enough to comment on, then it was good enough to upvote.
If a post or comment is particularly well thought-out, well-reasoned, or otherwise showing an understandable mastery of the issue at hand, it's worth considering upvoting it.
Don't upvote unless I'm absolutely confident, because I don't want to go skewing the statistics here, and I'm also pretty new at this.

My model for downvoting has been:

Don't do it until you know why other people do it (hence this post).

I've also been trying to understand why posts get comments and up/downvotes, but the two don't seem to correlate well. So are there different rules for upvoting comments versus posts?

[-]NancyLebovitz

10y60

Some good posts don't get many comments, perhaps because they've covered the subject thoroughly

I think posts are more likely to get a lot of comments if they're controversial or have the effect of encouraging people to talk about personal experience.

[-]Username

10y00

There's always going to be a difference between "what criteria for downvoting would make everyone better off" and "what criteria do people actually use". People will downvote to shut up opposing views, not because that's a good reason to downvote, but because dinging someone the -1 karma for an opposing view is more effective than not doing so and nobody's going to stop them.

[-]William_Quixote

10y20

I down vote posts I think are bad for the sites reputation / public image. I think people who like controversy or trolling privately benefit from such posts and externalize reputation all harms onto the site overall. I also think people in general benefit from a forum with good members and don't factor in long term reputational effects. I use karma to help people internalize these better.

[-]TheAncientGeek

10y00

Is civil, well argued disagreement bad for the site's reputation?

[-]Sabiola

10y30

It shouldn't be. It's one of the things I love about the site.

[-]someonewrongonthenet

10y20

My suggestion would be to downvote anything factually misleading, and to upvote anything for which you personally learned something from or had insight from. (Not agreement or disagreement). There are exceptions, as some threads (such as this one) are actually measuring agreement/disagreement, but that should be easy to determine.

I’m pretty sure I’m more proud of them than of all the work I did in high school put together.

I don't think that's not a good thing...there have been people in this community who attempted to drive out those they didn't like by downvoting everything they posted. The community would probably be better off without the downvote, since those who aught to use it do not and those who aught not to use it do. (Not to mention that some people unfortunately attach negative emotion to being downvoted, even though they haven't actually harmed anyone)

[-]Artaxerxes

10y10

People say that the rules are pretty general, but there was once a user that was banned for using the downvote in a way that people weren't comfortable with. I think it might be a sensible idea for mods to be very clear about what is off limits.

[-]Nornagest

10y50

There have been explicit mod statements against retributive downvoting (though they rely on a not-entirely-obvious reading of the content deletion policy): see here. We could probably use better and more obvious policy articles on the wiki, granted.

Aside from that, as long as you're not playing voting games with sockpuppets, writing scripts to automate voting, or spending hours of your time on delivering votes and nothing else, I'd say you're pretty safe. There's not a lot of policy because not a lot of policy is needed to regulate typical voting behavior; behave typically in your voting (i.e. vote manually, after reading content, and don't go looking for content to downvote) and you'll be fine.

[-]Artaxerxes

10y10

I think having a proper policy article on the wiki stating much what you've just outlined would be a good thing.

[-]TheAncientGeek

10y00

Having it somewhere more obvious, like the front page, would .be good as well.

[-]Viliam

10y10

there was once a user that was banned for using the downvote in a way that people weren't comfortable with

There was one such user in about ten years of LessWrong existence. (Okay, two user accounts, but most likely they belonged to the same person.) The rules as they are now seem sufficient for 99.9% of users to avoid this fate.

[-][anonymous]10y-10

If you don't feel a strong need to vote, then don't do it.

[-]Viliam

10y40

The more people would vote, the more difficult it would be for an individual to abuse the voting system.

[-][anonymous]10y00

How would an individual abuse the voting system, if the votes are determined by others?

[-]Viliam

10y10

If there is on average one comment per vote, then a single user, by systematically upvoting the comments they like and systematically downvoting the comments they dislike, would pretty much decide the outcome.

If there are on average ten comments per vote, then a single user could only partially influence whether a comment receives high or low total score.

It is similar to election. If in some country 90% of citizens don't vote, a party supported by 5% of people can gain a majority in a parliament. If 100% of citizens vote, a party supported by 5% is mostly harmless.

[-][anonymous]10y00

That sounds like how the system is meant to be used. One of the basic rules of trolling is that "Whatever can be used, can also be abused".

For example, guns can be used in self defense, and for "spraying graffiti" at the mall. Is the gun (that would be the voting system) to blame, or the person who pressed the trigger (the person making the vote?)

Don't get political - that was the best analogy I can come up with.

[-]Ixiel

10y-20

Down votes are also scarce, while upvotes are not. I for one, in the interest of symmetry, only upvote if I have down votes left. And as a poster, I would prefer others to do likewise. Remember there is no cash value to karma so the value to the poster of an upvote it's the same as the value of a down vote.

Moderation Log