I work for CEA, but the following views are my own. I donât have any plans to change Forum policy around which topics are permitted, discouraged, etc. This response is just my attempt to think through some considerations other EAs might want to make around this topic.
--
There were some things I liked about this post, but my comments here will mostly involve areas where I disagree with something. Still, criticism notwithstanding:
I appreciate the moves the post makes toward being considerate (the content note, the emphasis on not calling out individuals).
Two points from the post that I think are generally correct and somewhat underrated in debates around moderation policy: You canât please everyone, and power relations within particular spaces can look very different than power relations outside of those spaces. This also rang true (though I consider it a good thing for certain âgroupsâ to be disempowered in public discussion spaces):
There is a negative selection effect in that the more that a group is disempowered and could benefit from having its views being given more consideration, the less likely it is to have to power to make this happen.
The claim that we should not have âlimited discussionsâ is closing the barn door after the horse is already out. The EA Forum, like almost every other discussion space, has limits already. Even spaces that donât limit âworldlyâ topics may still have meta-limits on style/âdiscourse norms (no personal attacks, serious posts only, etc.). Aside from (maybe?) 4Chan, itâs hard to think of well-known discussion spaces that truly have no limits. For example, posts on the EA Forum:
Canât advocate the use of violence.
Are restricted in the types of criticism they can apply: âWe should remove Cause X from EA because its followers tend to smell badâ wouldnât get moderator approval, even if no individually smelly people were named.
--
While I donât fully agree with every claim in Making Discussions Inclusive, I appreciated the way that its authors didnât call for an outright ban on any particular form of speechâinstead, they highlighted the ways that speech permissions may influence other elements of group discussion, and noted that groups are making trade-offs when they figure out how to handle speech.
This post also mostly did this, but occasionally slipped into more absolute statements that donât quite square with reality (though I assume one is meant to read the full post while keeping the word âusuallyâ in mind, to insert in various places). An example:
We believe that someone is excluded to a greater degree when they are not allowed to share their sincerely held beliefs than when they are merely exposed to beliefs that they disagree with.
This seems simplistic. The reality of âexclusionâ depends on which beliefs are held, which beliefs are exposed, and the overall context of the conversation. Iâve seen conversations where someone shoehorned their âsincerely held beliefsâ into a discussion to which they werenât relevant, in such an odious way that many people who were strained on various resources (including âtimeâ and âpatienceâ) were effectively forced out. Perhaps banning the shoehorning user would have excluded them to a âgreater degreeâ, but their actions excluded a lot of people, even if to a âlesser degreeâ. Which outcome would have been worse? Itâs a complicated question.
Iâd argue that keeping things civil and on-topic is frequently less exclusionary than allowing total free expression, especially as conversations grow, because some ideas/âstyles are repellent to almost everyone. If someone insists on leaving multi-page comments with Caps Lock on in every conversation within a Facebook group, Iâd rather ask them to leave than ask the annoyed masses to grit their teeth and bear it.
This is an extreme example, of course, so Iâll use a real-world example from another discussion space I frequent: Reddit.
On the main Magic: The Gathering subreddit, conversations about a recent tournament winner (a non-binary person) were frequently interrupted by people with strong opinions about the pronoun âtheyâ being âconfusingâ or âweirdâ to use for a single person.
This is an intellectual position that may be worth discussing in other contexts, but in the context of these threads, it appeared hundreds of times and made it much more tedious to pick out actual Magic: The Gathering content. Within days, these users were being kicked out by moderators, and the forum became more readable as a result, to what Iâd guess was the collective relief of a large majority of users.
--
The general point Iâm trying to make:
âSomething nearly everyone dislikesâ is often going to be worth excluding even from the most popular, mainstream discussion venues.
In the context of EA, conversations that are genuinely about effective do-gooding should be protected, but I donât think several of your examples really fit that pattern:
Corruption in poor countries being caused by âcharacter flawsâ seems like a non sequitur.
When discussing ways to reduce corruption, we can talk about history, RCT results, and economic theoryâbut why personal characteristics?
Even if it were the case that people in Country A were somehow more âflawedâ than people in Country B, this only matters if it shows up in our data, and at that point, itâs just a set of facts about the world (e.g. âgovernment officials in A are more likely to demand bribes than officials in B, and bribery demands are inversely correlated with transfer impact, which means we should prefer to fund transfers in Bâ). I donât see the point of discussing the venality of the A-lish compared to the B-nians separately from actual data.
I think honest advocates for cash-transfer RCTs could quite truthfully state that they arenât trying to study whether poor people are âlazyâ. Someoneâs choice not to work doesnât have to be the target of criticism, even if it influences the estimated benefit of a cash transfer to that person. Itâs also possible to conclude that poor people discount the future without attaching the âcharacter flawâ label.
Frankly, labels like this tend to obscure discussion more than they help, by obscuring actual data and creating fake explanations (âpoor people donât care as much about the future, which is badâ < âpoor people donât care as much about the future, but this is moderated by factors A and B, and is economically rational if we factor in C, and hereâs a model for how we can encourage financial planning by people at different income levelsâ).
The same problem applies to your discussion of female influence and power; whether or not a personâs choices have led them to have less power seems immaterial to understanding which distributions of power tend to produce the best outcomes, and how particular policies might move us toward the best distributions.
To summarize the list of points above: In general, discussions of whether a state of the world is ârightâ, or whether a person is âgoodâ or âdeservingâ, donât make for great EA content. While I wouldnât prohibit them, I think they are far more tempting than they are useful, and that we should almost always try to use âif A, then Bâ reasoning rather than âhooray, B!â reasoning.
Of course, âthis reasoning style tends to be badâ doesnât mean âprohibit it entirelyâ. But it makes the consequence of limiting speech topics seem a bit less damaging, compared to what we could gain by being more inclusive. (Again, I donât actually think we should add more limits in any particular place, including the EA Forum. Iâm just pointing out considerations that other EAs might want to make when they think about these topics.)
âThe claim that we should not have âlimited discussionsâ is closing the barn door after the horse is already out.ââSome discussions about what should or should not allowed to be discussed are much more politicised than others and hence much more damaging. (In case it seems like Iâve contradicted myself here, my point is not that discussions about what should be allowed to be discussed should be banned, merely that a serious movement towards banning particular ideas encourages more of these adversarial discussions and hence the more positive and welcoming environment that is typically promised rarely materialises)
âWe believe that someone is excluded to a greater degree when they are not allowed to share their sincerely held beliefs than when they are merely exposed to beliefs that they disagree with.ââThis only referred to the degree of exclusion that would be experienced by a single individual for one of those two options. Obviously, the number of people affected is important as well as you point out.
Perhaps it is possible to discuss those issues while dodging political landmines, but thereâs also the worry about people being less willing to share views too close to the edge of the Overton Window.
I work for CEA, but the following views are my own. I donât have any plans to change Forum policy around which topics are permitted, discouraged, etc. This response is just my attempt to think through some considerations other EAs might want to make around this topic.
--
There were some things I liked about this post, but my comments here will mostly involve areas where I disagree with something. Still, criticism notwithstanding:
I appreciate the moves the post makes toward being considerate (the content note, the emphasis on not calling out individuals).
Two points from the post that I think are generally correct and somewhat underrated in debates around moderation policy: You canât please everyone, and power relations within particular spaces can look very different than power relations outside of those spaces. This also rang true (though I consider it a good thing for certain âgroupsâ to be disempowered in public discussion spaces):
The claim that we should not have âlimited discussionsâ is closing the barn door after the horse is already out. The EA Forum, like almost every other discussion space, has limits already. Even spaces that donât limit âworldlyâ topics may still have meta-limits on style/âdiscourse norms (no personal attacks, serious posts only, etc.). Aside from (maybe?) 4Chan, itâs hard to think of well-known discussion spaces that truly have no limits. For example, posts on the EA Forum:
Canât advocate the use of violence.
Are restricted in the types of criticism they can apply: âWe should remove Cause X from EA because its followers tend to smell badâ wouldnât get moderator approval, even if no individually smelly people were named.
--
While I donât fully agree with every claim in Making Discussions Inclusive, I appreciated the way that its authors didnât call for an outright ban on any particular form of speechâinstead, they highlighted the ways that speech permissions may influence other elements of group discussion, and noted that groups are making trade-offs when they figure out how to handle speech.
This post also mostly did this, but occasionally slipped into more absolute statements that donât quite square with reality (though I assume one is meant to read the full post while keeping the word âusuallyâ in mind, to insert in various places). An example:
We believe that someone is excluded to a greater degree when they are not allowed to share their sincerely held beliefs than when they are merely exposed to beliefs that they disagree with.
This seems simplistic. The reality of âexclusionâ depends on which beliefs are held, which beliefs are exposed, and the overall context of the conversation. Iâve seen conversations where someone shoehorned their âsincerely held beliefsâ into a discussion to which they werenât relevant, in such an odious way that many people who were strained on various resources (including âtimeâ and âpatienceâ) were effectively forced out. Perhaps banning the shoehorning user would have excluded them to a âgreater degreeâ, but their actions excluded a lot of people, even if to a âlesser degreeâ. Which outcome would have been worse? Itâs a complicated question.
Iâd argue that keeping things civil and on-topic is frequently less exclusionary than allowing total free expression, especially as conversations grow, because some ideas/âstyles are repellent to almost everyone. If someone insists on leaving multi-page comments with Caps Lock on in every conversation within a Facebook group, Iâd rather ask them to leave than ask the annoyed masses to grit their teeth and bear it.
This is an extreme example, of course, so Iâll use a real-world example from another discussion space I frequent: Reddit.
On the main Magic: The Gathering subreddit, conversations about a recent tournament winner (a non-binary person) were frequently interrupted by people with strong opinions about the pronoun âtheyâ being âconfusingâ or âweirdâ to use for a single person.
This is an intellectual position that may be worth discussing in other contexts, but in the context of these threads, it appeared hundreds of times and made it much more tedious to pick out actual Magic: The Gathering content. Within days, these users were being kicked out by moderators, and the forum became more readable as a result, to what Iâd guess was the collective relief of a large majority of users.
--
The general point Iâm trying to make:
âSomething nearly everyone dislikesâ is often going to be worth excluding even from the most popular, mainstream discussion venues.
In the context of EA, conversations that are genuinely about effective do-gooding should be protected, but I donât think several of your examples really fit that pattern:
Corruption in poor countries being caused by âcharacter flawsâ seems like a non sequitur.
When discussing ways to reduce corruption, we can talk about history, RCT results, and economic theoryâbut why personal characteristics?
Even if it were the case that people in Country A were somehow more âflawedâ than people in Country B, this only matters if it shows up in our data, and at that point, itâs just a set of facts about the world (e.g. âgovernment officials in A are more likely to demand bribes than officials in B, and bribery demands are inversely correlated with transfer impact, which means we should prefer to fund transfers in Bâ). I donât see the point of discussing the venality of the A-lish compared to the B-nians separately from actual data.
I think honest advocates for cash-transfer RCTs could quite truthfully state that they arenât trying to study whether poor people are âlazyâ. Someoneâs choice not to work doesnât have to be the target of criticism, even if it influences the estimated benefit of a cash transfer to that person. Itâs also possible to conclude that poor people discount the future without attaching the âcharacter flawâ label.
Frankly, labels like this tend to obscure discussion more than they help, by obscuring actual data and creating fake explanations (âpoor people donât care as much about the future, which is badâ < âpoor people donât care as much about the future, but this is moderated by factors A and B, and is economically rational if we factor in C, and hereâs a model for how we can encourage financial planning by people at different income levelsâ).
The same problem applies to your discussion of female influence and power; whether or not a personâs choices have led them to have less power seems immaterial to understanding which distributions of power tend to produce the best outcomes, and how particular policies might move us toward the best distributions.
To summarize the list of points above: In general, discussions of whether a state of the world is ârightâ, or whether a person is âgoodâ or âdeservingâ, donât make for great EA content. While I wouldnât prohibit them, I think they are far more tempting than they are useful, and that we should almost always try to use âif A, then Bâ reasoning rather than âhooray, B!â reasoning.
Of course, âthis reasoning style tends to be badâ doesnât mean âprohibit it entirelyâ. But it makes the consequence of limiting speech topics seem a bit less damaging, compared to what we could gain by being more inclusive. (Again, I donât actually think we should add more limits in any particular place, including the EA Forum. Iâm just pointing out considerations that other EAs might want to make when they think about these topics.)
âThe claim that we should not have âlimited discussionsâ is closing the barn door after the horse is already out.ââSome discussions about what should or should not allowed to be discussed are much more politicised than others and hence much more damaging. (In case it seems like Iâve contradicted myself here, my point is not that discussions about what should be allowed to be discussed should be banned, merely that a serious movement towards banning particular ideas encourages more of these adversarial discussions and hence the more positive and welcoming environment that is typically promised rarely materialises)
âWe believe that someone is excluded to a greater degree when they are not allowed to share their sincerely held beliefs than when they are merely exposed to beliefs that they disagree with.ââThis only referred to the degree of exclusion that would be experienced by a single individual for one of those two options. Obviously, the number of people affected is important as well as you point out.
Perhaps it is possible to discuss those issues while dodging political landmines, but thereâs also the worry about people being less willing to share views too close to the edge of the Overton Window.