An interesting thing that I and some other people on the LessWrong team noticed (as well as some users) was that since we created karma notifications we feel a lot more hesitant to downvote older comments, since we know that this will show up for the other users as a negative notification. I also feel a lot more hesitant to retract my own strong upvotes or upvotes in general since the author of the comment will see that as a downvote.
I’ve had many days in a row in which I received +20 or +30 karma, followed by a single day where by chance I received a single downvote and ended up at −2. The emotional valence of having a single day at −2 was somehow stronger than the emotional valence of multiple days of +20 or +30.
What I noticed on the EA forum is the whole karma thing is messing up with my S1 processes and makes me unhappy on average. I’ve not only turned off the notifications, but also hidden all karma displays in comments via css, and the experience is much better.
I… feel conflicted about people deactivating the display of karma on their own comments. In many ways karma (and downvotes in particular) serve as a really important feedback source, and I generally think that people who reliably get downvoted should change how they are commenting, and them not doing so usually comes at high cost. I think this is more relevant to new users, but is still relevant for most users.
Deactivating karma displays feels a bit to me like someone who shows up at a party and says “I am not going to listen to any subtle social feedback that people might give me about my behavior, and I will just do things until someone explicitly tells me to stop”, which I think is sometimes the correct behavior and has some good properties in terms of encouraging diversity of discussion, but I also expect that this can have some pretty large negative impact on the trust and quality of the social atmosphere.
On the other hand, I want people to have control over the incentives that they are under, and think it’s important to give users a lot of control over how they want to be influenced by the platform.
And there is also the additional thing, which is that if users just deactivate the karma display for their comments without telling anyone then that creates an environment of ambiguity where it’s very unclear whether someone receives the feedback you are giving them at all. In the party metaphor this would be like showing up and not telling anyone that you are not going to listen to subtle social feedback, which I think can easily lead to unnecessary escalation of conflict.
I don’t have a considered opinion on what to incentivize here, besides being pretty confident that I wouldn’t want most people to deactivate their karma displays, and that I am glad that you told me here that you did. This means that I will err on the side of leaving feedback by replying in addition to voting (though this obviously comes at a significant cost to me, so it might be game theoretically better for me to not shift towards replying, but I am not sure of that. Will think more about it).
There are also some common-knowledge effects that get really weird when one person is interacting with the discussion with a different set of data than I am seeing. I.e. I am going to reply to a downvoted comment in a way that assumes that many people thought the comment was bad and will try to explain potential reasons for why people might have downvoted it, but if you have karma displays disabled then you might perceive me as making a kind of social attack where I claim the support of some kind of social group without backing it up. I think this makes me quite hesitant to participate in discussions with that kind of weird information asymmetry.
Well… you can’t actually stop people from activating custom CSS that hides karma values. It doesn’t matter how you feel about it—you can’t affect it! It’s therefore probably best to create some mechanism that gives people what they want to get out of hiding karma, while still giving you what you want out of showing people karma (e.g., a “hide karma but give me a notification if one of my comments is quite strongly downvoted” option—not suggesting this exact thing, just brainstorming…).
Hmm, I agree that I can’t prevent it in that sense, but I think defaults matter a lot here, as does just normal social feedback and whatever the social norms are.
It’s not at all clear to me that the current equilibrium isn’t pretty decent, where people can do it, but it’s reasonably inconvenient to do it, and so allows the people who are disproportionately negatively affected by karma notification to go that route. I would be curious in whether there are any others who do the same as Jan does, and if there are many, then we can figure out what the common motivations are and see whether it makes sense to elevate it to some site-level feature.
It’s not at all clear to me that the current equilibrium isn’t pretty decent, where people can do it, but it’s reasonably inconvenient to do it, and so allows the people who are disproportionately negatively affected by karma notification to go that route.
But this is an extremely fragile equilibrium. It can be broken by, say, someone posting a set of simple instructions on how to do this. For instance:
Anyone running the uBlock Origin browser extension can append several lines to their “My Filters” tab in the uBlock extension preferences, and thus totally hide all karma-related UI elements on Less Wrong. (PM me if you want the specific lines to append.)
Or someone makes a browser extension to do this. Or a user style. Or…
FWIW I also think it’s quite possible the current equilibrium is decent (which is part of reasons why I did not posted something like “How did I turned karma off” with simple instruction about how to do it on the forum, which I did consider). On the other hand I’d be curious about more people trying it and reporting their experiences.
I suspect many people kind of don’t have this action in the space of things they usually consider—I’d expect what most people would do is 1) just stop posting 2) write about their negative experience 3) complain privately.
Actually I turned the karma for all comments, not just mine. The bold claim is my individual taste in what’s good on the EA forum is in important ways better than the karma system, and the karma signal is similar to sounds made by a noisy mob. If I want I can actually predict what average sounds will the crowd make reasonably well, so it is not any new source of information. But it still messes up with your S1 processing and motivations.
Continuing with the party metaphor, I think it is generally not that difficult to understand what sort of behaviour will make you popular at a party, and what sort of behaviours even when they are quite good in a broader scheme of things will make you unpopular at parties. Also personally I often feel something like “I actually want to have good conversations about juicy topics in a quite place, unfortunately you all people are congregating at this super loud space, with all these status games, social signals, and ethically problematic norms how to treat other people” toward most parties.
Overall I posted this here because it seemed like an interesting datapoint. Generally I think it would be great if people moved toward writing information rich feedback instead of voting, so such shift seems good. From what I’ve seen on EA forum it’s quite rarely “many people” doing anything. More often it is like 6 users upvote a comment, 1user strongly downvotes it, something like karma 2 is a result. I would guess you may be in larger risk of distorted perception that this represents some meaningful opinion of the community. (Also I see some important practical cases where people are misled by “noises of the crowd” and it influences them in a harmful way.)
If people are checking karma changes constantly and getting emotional validation or pain from the result, that seems like a bad result. And yes, the whole ‘one −2 and three +17s feels like everyone hates me’ thing is real, can confirm.
Because of the way we do batching you can’t check karma changes constantly (unless you go out of your way to change your setting) because we batch karma notifications on a 24h basis by default.
True. We did very intentionally avoid putting your total karma on the frontpage anywhere as most other platforms do to avoid people getting sucked into that unintentionally, but it you can still do that on your profile.
I hope we aren’t wasting a lot of people’s time by causing them to check their profile all the time. If we do, it might be the correct choice to also only update that number every 24h.
Do our karma karma notifications disappear if you don’t check them that day? My model of Zvi suggested to me this is attention-grabbing and bad. I wonder if it’s better to let folks be notified of all days’ karma updates ’til their most recent check in, and maybe also see all historical ones ordered by date if they click on a further button, so that the info isn’t lost and doesn’t feel scarce.
Which is definitely better than it expiring, and 24h batching is better than instantaneous feedback (unless you were going to check posts individually for information already, in which case things are already quite bad). It’s not obvious to me what encouraging daily checks here is doing for discourse as opposed to being a Skinner box.
The motivation was (among other things) several people saying to us “yo, I wish LessWrong was a bit more of a skinner box because right now it’s so throughly not a skinner box that it just doesn’t make it into my habits, and I endorse it being a stronger habit than it currently is.”
It’s interesting to see how people’s votes on a post or comment are affected by other comments. I’ve noticed that a burst of vote count changes often appears after a new and apparently influential reply shows up.
Yeah, I had the same occurrence + feeling recently when I wrote the quant trading post. It felt like: “Wait, who would downvote this post...??” It’s probably more likely that someone just retracted an upvote.
Reminder: If a person is not willing to explain their voting decisions, you are under no obligation to waste cognition trying to figure them out. They don’t deserve that. They probably don’t even want that.
That depends on what norm is in place. If the norm is to explain downvoting, then people should explain, otherwise there is no issue in not doing so. So the claim you are making is that the norm should be for people to explain. The well-known counterargument is that this disincentivizes downvoting.
you are under no obligation to waste cognition trying to figure them out
There is rarely an obligation to understand things, but healthy curiosity ensures progress on recurring events, irrespective of morality of their origin. If an obligation would force you to actually waste cognition, don’t accept it!
So the claim you are making is that the norm should be for people to explain
I’m not really making that claim. A person doesn’t have to do anything condemnable to be in a state of not deserving something. If I don’t pay the baker, I don’t deserve a bun. I am fine with not deserving a bun, as I have already eaten.
The baker shouldn’t feel like I am owed a bun.
Another metaphor is that the person who is beaten on the street by silent, masked assailants should not feel like they owe their oppressors an apology.
Do you mean anything by this beyond “you don’t have an obligation to figure out why people voted one way or another, period”? (Or do you think that I [i.e., the general Less Wrong commenter] do have such an obligation?)
Edit: Also, the “They don’t deserve that” bit confuses me. Are you suggesting that understanding why people upvoted or downvoted your comment is a favor that you are doing for them?
Sometimes a person wont want to reply and say outright that they thought the comment was bad, because it’s just not pleasant, and perhaps not necessary. Instead, they might just reply with information that they think you might be missing, which you could use to improve, if you chose to. With them, an engaged interlocutor will be able to figure out what isn’t being said. With them, it can be productive to try to read between the lines.
Are you suggesting that understanding why people upvoted or downvoted your comment is a favor that you are doing for them?
Isn’t everything relating to writing good comments a favor, that you are doing for others. But I don’t really think in terms of favors. All I mean to say is that we should write our comments for the sorts of people who give feedback. Those are the good people. Those are the people who’re a part of a good faith self-improving discourse. Their outgroup are maybe not so good, and we probably shouldn’t try to write for their sake.
I think I disagree. If you are getting downvoted by 5 people and one of them explains why, then even if the other 4 are not explaining their reasoning it’s often reasonable to assume that more than just the one person had the same complaints, and as such you likely want to update more that it’s better for you to change what you are doing.
Thoughts on negative karma notifications:
An interesting thing that I and some other people on the LessWrong team noticed (as well as some users) was that since we created karma notifications we feel a lot more hesitant to downvote older comments, since we know that this will show up for the other users as a negative notification. I also feel a lot more hesitant to retract my own strong upvotes or upvotes in general since the author of the comment will see that as a downvote.
I’ve had many days in a row in which I received +20 or +30 karma, followed by a single day where by chance I received a single downvote and ended up at −2. The emotional valence of having a single day at −2 was somehow stronger than the emotional valence of multiple days of +20 or +30.
What I noticed on the EA forum is the whole karma thing is messing up with my S1 processes and makes me unhappy on average. I’ve not only turned off the notifications, but also hidden all karma displays in comments via css, and the experience is much better.
I… feel conflicted about people deactivating the display of karma on their own comments. In many ways karma (and downvotes in particular) serve as a really important feedback source, and I generally think that people who reliably get downvoted should change how they are commenting, and them not doing so usually comes at high cost. I think this is more relevant to new users, but is still relevant for most users.
Deactivating karma displays feels a bit to me like someone who shows up at a party and says “I am not going to listen to any subtle social feedback that people might give me about my behavior, and I will just do things until someone explicitly tells me to stop”, which I think is sometimes the correct behavior and has some good properties in terms of encouraging diversity of discussion, but I also expect that this can have some pretty large negative impact on the trust and quality of the social atmosphere.
On the other hand, I want people to have control over the incentives that they are under, and think it’s important to give users a lot of control over how they want to be influenced by the platform.
And there is also the additional thing, which is that if users just deactivate the karma display for their comments without telling anyone then that creates an environment of ambiguity where it’s very unclear whether someone receives the feedback you are giving them at all. In the party metaphor this would be like showing up and not telling anyone that you are not going to listen to subtle social feedback, which I think can easily lead to unnecessary escalation of conflict.
I don’t have a considered opinion on what to incentivize here, besides being pretty confident that I wouldn’t want most people to deactivate their karma displays, and that I am glad that you told me here that you did. This means that I will err on the side of leaving feedback by replying in addition to voting (though this obviously comes at a significant cost to me, so it might be game theoretically better for me to not shift towards replying, but I am not sure of that. Will think more about it).
There are also some common-knowledge effects that get really weird when one person is interacting with the discussion with a different set of data than I am seeing. I.e. I am going to reply to a downvoted comment in a way that assumes that many people thought the comment was bad and will try to explain potential reasons for why people might have downvoted it, but if you have karma displays disabled then you might perceive me as making a kind of social attack where I claim the support of some kind of social group without backing it up. I think this makes me quite hesitant to participate in discussions with that kind of weird information asymmetry.
Well… you can’t actually stop people from activating custom CSS that hides karma values. It doesn’t matter how you feel about it—you can’t affect it! It’s therefore probably best to create some mechanism that gives people what they want to get out of hiding karma, while still giving you what you want out of showing people karma (e.g., a “hide karma but give me a notification if one of my comments is quite strongly downvoted” option—not suggesting this exact thing, just brainstorming…).
Hmm, I agree that I can’t prevent it in that sense, but I think defaults matter a lot here, as does just normal social feedback and whatever the social norms are.
It’s not at all clear to me that the current equilibrium isn’t pretty decent, where people can do it, but it’s reasonably inconvenient to do it, and so allows the people who are disproportionately negatively affected by karma notification to go that route. I would be curious in whether there are any others who do the same as Jan does, and if there are many, then we can figure out what the common motivations are and see whether it makes sense to elevate it to some site-level feature.
But this is an extremely fragile equilibrium. It can be broken by, say, someone posting a set of simple instructions on how to do this. For instance:
Anyone running the uBlock Origin browser extension can append several lines to their “My Filters” tab in the uBlock extension preferences, and thus totally hide all karma-related UI elements on Less Wrong. (PM me if you want the specific lines to append.)
Or someone makes a browser extension to do this. Or a user style. Or…
FWIW I also think it’s quite possible the current equilibrium is decent (which is part of reasons why I did not posted something like “How did I turned karma off” with simple instruction about how to do it on the forum, which I did consider). On the other hand I’d be curious about more people trying it and reporting their experiences.
I suspect many people kind of don’t have this action in the space of things they usually consider—I’d expect what most people would do is 1) just stop posting 2) write about their negative experience 3) complain privately.
Actually I turned the karma for all comments, not just mine. The bold claim is my individual taste in what’s good on the EA forum is in important ways better than the karma system, and the karma signal is similar to sounds made by a noisy mob. If I want I can actually predict what average sounds will the crowd make reasonably well, so it is not any new source of information. But it still messes up with your S1 processing and motivations.
Continuing with the party metaphor, I think it is generally not that difficult to understand what sort of behaviour will make you popular at a party, and what sort of behaviours even when they are quite good in a broader scheme of things will make you unpopular at parties. Also personally I often feel something like “I actually want to have good conversations about juicy topics in a quite place, unfortunately you all people are congregating at this super loud space, with all these status games, social signals, and ethically problematic norms how to treat other people” toward most parties.
Overall I posted this here because it seemed like an interesting datapoint. Generally I think it would be great if people moved toward writing information rich feedback instead of voting, so such shift seems good. From what I’ve seen on EA forum it’s quite rarely “many people” doing anything. More often it is like 6 users upvote a comment, 1user strongly downvotes it, something like karma 2 is a result. I would guess you may be in larger risk of distorted perception that this represents some meaningful opinion of the community. (Also I see some important practical cases where people are misled by “noises of the crowd” and it influences them in a harmful way.)
If people are checking karma changes constantly and getting emotional validation or pain from the result, that seems like a bad result. And yes, the whole ‘one −2 and three +17s feels like everyone hates me’ thing is real, can confirm.
Because of the way we do batching you can’t check karma changes constantly (unless you go out of your way to change your setting) because we batch karma notifications on a 24h basis by default.
I mean, you can definitely check your karma multiple times a day to see where the last two sig digits are at, which is something I sometimes do.
True. We did very intentionally avoid putting your total karma on the frontpage anywhere as most other platforms do to avoid people getting sucked into that unintentionally, but it you can still do that on your profile.
I hope we aren’t wasting a lot of people’s time by causing them to check their profile all the time. If we do, it might be the correct choice to also only update that number every 24h.
I’ve never checked my karma total on LW 2.0 to see how it’s changed.
In my case, it sure feels like I check my karma often because I often want to know what my karma is, but maybe others differ.
Do our karma karma notifications disappear if you don’t check them that day? My model of Zvi suggested to me this is attention-grabbing and bad. I wonder if it’s better to let folks be notified of all days’ karma updates ’til their most recent check in, and maybe also see all historical ones ordered by date if they click on a further button, so that the info isn’t lost and doesn’t feel scarce.
Nah, they accumulate until you click on them.
Which is definitely better than it expiring, and 24h batching is better than instantaneous feedback (unless you were going to check posts individually for information already, in which case things are already quite bad). It’s not obvious to me what encouraging daily checks here is doing for discourse as opposed to being a Skinner box.
The motivation was (among other things) several people saying to us “yo, I wish LessWrong was a bit more of a skinner box because right now it’s so throughly not a skinner box that it just doesn’t make it into my habits, and I endorse it being a stronger habit than it currently is.”
See this comment and thread.
It’s interesting to see how people’s votes on a post or comment are affected by other comments. I’ve noticed that a burst of vote count changes often appears after a new and apparently influential reply shows up.
Yeah, I had the same occurrence + feeling recently when I wrote the quant trading post. It felt like: “Wait, who would downvote this post...??” It’s probably more likely that someone just retracted an upvote.
Reminder: If a person is not willing to explain their voting decisions, you are under no obligation to waste cognition trying to figure them out. They don’t deserve that. They probably don’t even want that.
That depends on what norm is in place. If the norm is to explain downvoting, then people should explain, otherwise there is no issue in not doing so. So the claim you are making is that the norm should be for people to explain. The well-known counterargument is that this disincentivizes downvoting.
There is rarely an obligation to understand things, but healthy curiosity ensures progress on recurring events, irrespective of morality of their origin. If an obligation would force you to actually waste cognition, don’t accept it!
I’m not really making that claim. A person doesn’t have to do anything condemnable to be in a state of not deserving something. If I don’t pay the baker, I don’t deserve a bun. I am fine with not deserving a bun, as I have already eaten.
The baker shouldn’t feel like I am owed a bun.
Another metaphor is that the person who is beaten on the street by silent, masked assailants should not feel like they owe their oppressors an apology.
Do you mean anything by this beyond “you don’t have an obligation to figure out why people voted one way or another, period”? (Or do you think that I [i.e., the general Less Wrong commenter] do have such an obligation?)
Edit: Also, the “They don’t deserve that” bit confuses me. Are you suggesting that understanding why people upvoted or downvoted your comment is a favor that you are doing for them?
Sometimes a person wont want to reply and say outright that they thought the comment was bad, because it’s just not pleasant, and perhaps not necessary. Instead, they might just reply with information that they think you might be missing, which you could use to improve, if you chose to. With them, an engaged interlocutor will be able to figure out what isn’t being said. With them, it can be productive to try to read between the lines.
Isn’t everything relating to writing good comments a favor, that you are doing for others. But I don’t really think in terms of favors. All I mean to say is that we should write our comments for the sorts of people who give feedback. Those are the good people. Those are the people who’re a part of a good faith self-improving discourse. Their outgroup are maybe not so good, and we probably shouldn’t try to write for their sake.
I think I disagree. If you are getting downvoted by 5 people and one of them explains why, then even if the other 4 are not explaining their reasoning it’s often reasonable to assume that more than just the one person had the same complaints, and as such you likely want to update more that it’s better for you to change what you are doing.
We don’t disagree.
Cool