Lots of meta-level thoughts. I again apologize for not participating on the object-level. My life currently feels extremely busy and I’ve been stressing out about a bunch of other stuff that has prevented me from engaging with a lot of this.
Because the opinions of people are the best pointer I currently have, without spending a lot of time writing things myself, here is roughly where I stand in terms of what I agree and disagree with. I would usually prefer to just explain my own position, but because time is short and I am the closest to a central decision-maker that LessWrong has, it seems particularly important for people to have a model of where I am coming from.
I think Ruby said a bunch of correct things, in particular in the original proposed norms document. I think it was quite a good choice to make a concrete set of norms, and felt like the discussion on that was pretty useful.
I think this is the core part of Ruby’s post that I had the strongest reaction to:
I applaud the efforts of all those trying to do good, those who donate their time, attention, and money towards making this world better. Unfortunately, good intentions don’t always cause good outcomes. I regret to say that after careful investigation I believe cause X is in fact harmful (as I will elaborate), and that those who have supported it should place their efforts elsewhere. It is important that we pay attention to factors A and B . . .
As both Ray and Zvi pointed out, I think this as the norm would be quite exceptionally bad and I would hate to have to participate in a forum in which this is the norm (and the fact that this kinda feels like the norm on the EA Forum is quite bad and makes me a lot less interested in participating. See this comment as an example of this norm being enforced, which I strongly downvoted). I think this as the norm can make some sense if you are in a context that’s kind of like the UN, where if you accidentally annoy someone war might happen, and the cultural barriers + language barriers prevent almost any form of long-term trust to happen, but is a very bad choice in most environments, including LW.
I do also think version 2 is a bad idea, though I do think it’s less of a bad idea than the first one, mostly because of the reasons that Benquo outlined in Blatant Lies are the Best Kind!. The second version feels like its at least something I can call out as wrong. The first version feels like something where in order to call it out I would have to construct a whole ontology of incentives on the fly, which would inevitably cause me to lose the interest of the audience, and feels much harder to deal with.
I think the disagreement between version 1 and version 2 is maybe what all the rest of the discussion is about, but I don’t really know. I had some sense of people talking past each other in ways that wasn’t super useful.
I feel like the correct plan at this point in the situation is to make guesses at potential norms and rules we could have and curiously inspect what goes wrong if we decide to follow them. I think we are pretty far from having a concrete set of norms we can implement more widely on LessWrong, but feel optimistic about our ability to make progress on things.
In the spirit of pointing out incentives, I will say straightforwardly that I am somewhat afraid of both Benquo and Jessica in this conversation and so feel a lot of non-endorsed feelings of wanting to appease you and not get into conflict with you. I don’t intend to act on those feelings, but it’s definitely framing this discussion a bunch for me and I expect will cause me to sometimes say things I don’t endorse.
The three Versions have been subject of greater attention than I expected, I suppose examples are fairly memorable and easy to point at. For clarity, their main purpose was not to exemplify norms, but merely to demonstrate the grammatical concept of “core message” vs “extra information” with positively- and negatively-valenced examples. This is why they were so over the top—I wanted it to be unmistakable to even the most “tone-deaf” reader that the imagined author was saying a lot more than the putative propositional object-level thesis.
By which I mean to say, none of the three Versions encapsulate the norms I currently estimate are best, and in my mind, the discussion hasn’t been about arguing for one over the others, e.g. 1 vs 2. Afterall, they’re not norms to be argued for (though they help in describing norms).
However, clarifying my position with reference to the three Versions:
Most of the time, on LW, I’d want communication to look like Version 3 where nobody’s putting anything much positive/negative in the side channels which isn’t also about the core message. Instead thoughts are stated clearly and plainly without decoration.
[I edited this one after reflection.] I think that the extremely politics-y Version 1 is not something I want to see much in my ideal world, and not much on my ideal LessWrong. It’s not something I’d push people to do. Not like that. That said, some topics are very fraught and I think it’s better to discuss them with clear honest signalling of intent that is as rich as Version 1 than not at all or in a way that is highly destructive.
I can read Version 1 as being said by someone very politics-y, but also by someone being very compassionate. I could see that approach being key to enable important and productive discussions that couldn’t happen otherwise, e.g. where political tensions are already tight.
I do think there is a “steel” Version 1 which is in commonly virtuous but probably supererogatory. It’s not something I think I’d want to enforce in any way, but I might encourage it. By this I mean something like it is virtuous for an individual to put some effort into appearing non-X when they don’t actually mean to be X, and being perceived as X is bad. X could be hostile/dismissive/demeaning/judgmental/threatening/etc..
I suspect even a much milder version of Version 1 would seem bad to many here. Going forward, I want to dig into what “steel Version 1” would look like and whether it’s in fact virtuous/even assuming it was, whether it’s good to encourage it. Plus when even full-blown Version 1 is actually a good idea too.
[Further edit] Admittedly, I also have a strong prior that it’s a mistake most of the time to be hostile/dismissive/demeaning/judgmental/threatening/etc. on LessWrong and would encourage people to update towards it probably not being right to relate that way to others.
I don’t think I can honestly accept mere explicit endorsements of a high-level opinion here, because as far as I can tell, such endorsements are often accompanied by behavior, or other endorsements, that seem (to me) to contradict them. I guess that’s why the examples have attracted so much attention—I have more of an expectation that they correspond to the intuitions people will make decisions with, and those are what I want to be arguing with.
Having written that, it occurs to me that I too could do a better job giving examples that illustrate the core considerations I’m trying to draw attention to, so I’ll make an effort to do that in the future.
Lots of meta-level thoughts. I again apologize for not participating on the object-level. My life currently feels extremely busy and I’ve been stressing out about a bunch of other stuff that has prevented me from engaging with a lot of this.
Because the opinions of people are the best pointer I currently have, without spending a lot of time writing things myself, here is roughly where I stand in terms of what I agree and disagree with. I would usually prefer to just explain my own position, but because time is short and I am the closest to a central decision-maker that LessWrong has, it seems particularly important for people to have a model of where I am coming from.
I think Ruby said a bunch of correct things, in particular in the original proposed norms document. I think it was quite a good choice to make a concrete set of norms, and felt like the discussion on that was pretty useful.
I think this is the core part of Ruby’s post that I had the strongest reaction to:
As both Ray and Zvi pointed out, I think this as the norm would be quite exceptionally bad and I would hate to have to participate in a forum in which this is the norm (and the fact that this kinda feels like the norm on the EA Forum is quite bad and makes me a lot less interested in participating. See this comment as an example of this norm being enforced, which I strongly downvoted). I think this as the norm can make some sense if you are in a context that’s kind of like the UN, where if you accidentally annoy someone war might happen, and the cultural barriers + language barriers prevent almost any form of long-term trust to happen, but is a very bad choice in most environments, including LW.
I do also think version 2 is a bad idea, though I do think it’s less of a bad idea than the first one, mostly because of the reasons that Benquo outlined in Blatant Lies are the Best Kind!. The second version feels like its at least something I can call out as wrong. The first version feels like something where in order to call it out I would have to construct a whole ontology of incentives on the fly, which would inevitably cause me to lose the interest of the audience, and feels much harder to deal with.
I think the disagreement between version 1 and version 2 is maybe what all the rest of the discussion is about, but I don’t really know. I had some sense of people talking past each other in ways that wasn’t super useful.
I feel like the correct plan at this point in the situation is to make guesses at potential norms and rules we could have and curiously inspect what goes wrong if we decide to follow them. I think we are pretty far from having a concrete set of norms we can implement more widely on LessWrong, but feel optimistic about our ability to make progress on things.
In the spirit of pointing out incentives, I will say straightforwardly that I am somewhat afraid of both Benquo and Jessica in this conversation and so feel a lot of non-endorsed feelings of wanting to appease you and not get into conflict with you. I don’t intend to act on those feelings, but it’s definitely framing this discussion a bunch for me and I expect will cause me to sometimes say things I don’t endorse.
The three Versions have been subject of greater attention than I expected, I suppose examples are fairly memorable and easy to point at. For clarity, their main purpose was not to exemplify norms, but merely to demonstrate the grammatical concept of “core message” vs “extra information” with positively- and negatively-valenced examples. This is why they were so over the top—I wanted it to be unmistakable to even the most “tone-deaf” reader that the imagined author was saying a lot more than the putative propositional object-level thesis.
By which I mean to say, none of the three Versions encapsulate the norms I currently estimate are best, and in my mind, the discussion hasn’t been about arguing for one over the others, e.g. 1 vs 2. Afterall, they’re not norms to be argued for (though they help in describing norms).
However, clarifying my position with reference to the three Versions:
Most of the time, on LW, I’d want communication to look like Version 3 where nobody’s putting anything much positive/negative in the side channels which isn’t also about the core message. Instead thoughts are stated clearly and plainly without decoration.
[I edited this one after reflection.] I think that the extremely politics-y Version 1 is not something I want to see much in my ideal world, and not much on my ideal LessWrong. It’s not something I’d push people to do. Not like that. That said, some topics are very fraught and I think it’s better to discuss them with clear honest signalling of intent that is as rich as Version 1 than not at all or in a way that is highly destructive.
I can read Version 1 as being said by someone very politics-y, but also by someone being very compassionate. I could see that approach being key to enable important and productive discussions that couldn’t happen otherwise, e.g. where political tensions are already tight.
I do think there is a “steel” Version 1 which is in commonly virtuous but probably supererogatory. It’s not something I think I’d want to enforce in any way, but I might encourage it. By this I mean something like it is virtuous for an individual to put some effort into appearing non-X when they don’t actually mean to be X, and being perceived as X is bad. X could be hostile/dismissive/demeaning/judgmental/threatening/etc..
I suspect even a much milder version of Version 1 would seem bad to many here. Going forward, I want to dig into what “steel Version 1” would look like and whether it’s in fact virtuous/even assuming it was, whether it’s good to encourage it. Plus when even full-blown Version 1 is actually a good idea too.
[Further edit] Admittedly, I also have a strong prior that it’s a mistake most of the time to be hostile/dismissive/demeaning/judgmental/threatening/etc. on LessWrong and would encourage people to update towards it probably not being right to relate that way to others.
I don’t think I can honestly accept mere explicit endorsements of a high-level opinion here, because as far as I can tell, such endorsements are often accompanied by behavior, or other endorsements, that seem (to me) to contradict them. I guess that’s why the examples have attracted so much attention—I have more of an expectation that they correspond to the intuitions people will make decisions with, and those are what I want to be arguing with.
Having written that, it occurs to me that I too could do a better job giving examples that illustrate the core considerations I’m trying to draw attention to, so I’ll make an effort to do that in the future.