(Copying across some comments I made on Facebook which are relevant to this.)
Thanks for the passionate feedback everyone. Whilst I don’t agree with all of the comments, I’m sorry for the mistakes I made. Since some of the comments above make similar comments, I’ll try to give general replies in some main-thread comments. I’ll also be reaching out to some of the people in the thread above to try to work out the best way forward.
My understanding is that the main worry that people have is about calling it the Effective Altruism Handbook vs. CEA’s Guide to Effective Altruism or similar. For the reasons given in my reply to Scott above, I think that calling it the EA Handbook is not a significant change from before: unless we ask Ryan to take down the old handbook, then whatever happens, there will be a CEA-selected resource called the EA Handbook. For reasons given above and below, I think that the new version of the Handbook is better than the old. I think that there is some value in explicitly replacing the old version for this reason, and since “EA Handbook” is a cleaner name. However, I do also get people’s worries about this being taken to represent the EA community as a whole. For that reason, I will make sure that the title page and introduction make clear that this is a project of CEA, and I will make clear in the introduction that others in the community would have selected different essays.
My preferred approach would then be to engage with people who have expressed concern, and see if there are changes we can make that alleviate their concerns (such as those we already plan to make based on Scott’s comment). If it appears that we can alleviate most of those concerns whilst retaining the value of the Handbook from from CEA’s perspective, it might be best to call it the Centre for Effective Altruism’s EA Handbook. Otherwise, we would rebrand. I’d be interested to hear in comments whether there are specific changes (articles to add/take away/design things) that would reassure you about this being called the EA Handbook.
In this comment I’ll reply to some of the more object-level criticisms. I want to apologize for how this seemed to others, but also give a clearer sense of our intentions. I think that it might seem that CEA has tried merely to push AI safety as the only thing to work on. We don’t think that, and that wasn’t our intention. Obviously, poorly realized intentions are still a problem, but I want to reassure people about CEA’s approach to these issues.
First, re there not being enough discussion of portfolios/comparative advantage, this is mentioned in two of the articles (“Prospecting for Gold” and “What Does (and Doesn’t) AI Mean for Effective Altruism?”). However, I think that we could have emphasised this more, and I will see if it’s possible to include a full article on coordination and comparative advantage.
Second, I’d like to apologise for the way the animal and global health articles came across. Those articles were commissioned at the same time as the long-term future article, and they share a common structure: What’s the case for this cause? What are some common concerns about that cause? Why might you choose not to support this cause? The intention was to show how many assumptions underlie a decision to focus on any cause, and to map out some of the debate between the different cause areas, rather than to illicitly push the long-term future. It looks like this didn’t come across, sorry. We didn’t initially commission sub-cause profiles on government, AI and biosecurity, which explains why those more specific articles follow a different structure (mostly talks given at EA Global).
Third, I want to explain some of the reasoning behind including several articles on AI. AI risk is a more unusual area, which is more susceptible to misinterpretation than global health or animal welfare. Partly for this reason, we thought that it was sensible to include several articles on this topic, with the intention that this would provide more needed background and convey more of the nuance of the idea. I will talk with some of the commenters above to discuss if it makes sense to do some sort of merge so that AI dominates the contents page less.
What about the possibility the Centre for Effective Altruism represents the community by editing the EA Handbook to reflect what the community values in spite of what the CEA concludes, the CEA excludes evaluations from the EA Handbook from which it currently diverges from the community, and it’s still called the ‘EA Handbook’ instead of ‘CEA’s Guide to EA?’ Obviously this wouldn’t carry EA forward with what the CEA thinks is maximum fidelity, but it’s clear many think the CEA is trying to spread the EA message with infidelity, while acting as though they’re the only actor in the movement others can trust to carry that message. It looks not only hypocritical but undermines faith in the CEA.
Altering the handbook so it’s more of a compromise between multiple actors in EA will redeem the reputation of the CEA. Without that, the CEA can’t carry EA forward with the fidelity at all, because the rest of the movement wouldn’t cooperate with them. In the meantime, the CEA and everyone else can hammer out what we think is the most good here on the EA Forum. If broader conclusions are drawn which line up with the CEA’s evaluation based on a consensus the CEA had behind their perspective the best arguments, that can be included in the next edition of the EA Handbook. Again, from the CEA’s perspective, that might seem like deliberately compromising the fidelity of EA in the short-term to appease others. But again, from the perspective of the CEA’s current critics, why they’re criticizing the 2nd edition of the EA Handbook is because they perceive themselves as protecting the fidelity of EA from the Centre for Effective Altruism. This could solve other contentious issues in EA, such as consideration of both s-risks and x-risks from AI. The EA Handbook could be published as close to identically as possible in multiple languages, which would prevent the CEA from selling EA one way in English, the EAF selling it another way in German, and creating more trust issues which would down the road just become sources of conflict, not unlike the criticism the EA Handbook, 2nd edition, is receiving now. Ultimately, this would be the CEA making a relatively short-term compromise to ensure the long-term fidelity of EA by demonstrating themselves as delegate and representative agency the EA community can still have confidence in.
Thanks for the comments Evan. First, I want to apologize for not seeking broader consultation earlier. This was clearly a mistake.
My plan now is to do as you suggest: talk to other actors in EA and get their feedback on what to include etc. Obviously any compromise is going to leave some unhappy—different groups do just favour different presentations of EA, so it seems unlikely to me that we will get a fully independent presentation that will please everyone. I also worry that democracy is not well suited to editorial decisions, and that the “electorate” of EA is ill-defined. If the full compromise approach fails, I think it would be best to release a CEA-branded resource which incorporates most of the feedback above. This option also seems to me to be cooperative, and to avoid harm to the fidelity of EA’s message, but I might be missing something.
Thanks for responding Max. I agree consulting some key actors but not going through a democratic makes sense. I appreciate you being able to respond to and incorporate all the feedback you’re receiving so quickly.
(Copying across some comments I made on Facebook which are relevant to this.)
Thanks for the passionate feedback everyone. Whilst I don’t agree with all of the comments, I’m sorry for the mistakes I made. Since some of the comments above make similar comments, I’ll try to give general replies in some main-thread comments. I’ll also be reaching out to some of the people in the thread above to try to work out the best way forward.
My understanding is that the main worry that people have is about calling it the Effective Altruism Handbook vs. CEA’s Guide to Effective Altruism or similar. For the reasons given in my reply to Scott above, I think that calling it the EA Handbook is not a significant change from before: unless we ask Ryan to take down the old handbook, then whatever happens, there will be a CEA-selected resource called the EA Handbook. For reasons given above and below, I think that the new version of the Handbook is better than the old. I think that there is some value in explicitly replacing the old version for this reason, and since “EA Handbook” is a cleaner name. However, I do also get people’s worries about this being taken to represent the EA community as a whole. For that reason, I will make sure that the title page and introduction make clear that this is a project of CEA, and I will make clear in the introduction that others in the community would have selected different essays.
My preferred approach would then be to engage with people who have expressed concern, and see if there are changes we can make that alleviate their concerns (such as those we already plan to make based on Scott’s comment). If it appears that we can alleviate most of those concerns whilst retaining the value of the Handbook from from CEA’s perspective, it might be best to call it the Centre for Effective Altruism’s EA Handbook. Otherwise, we would rebrand. I’d be interested to hear in comments whether there are specific changes (articles to add/take away/design things) that would reassure you about this being called the EA Handbook.
In this comment I’ll reply to some of the more object-level criticisms. I want to apologize for how this seemed to others, but also give a clearer sense of our intentions. I think that it might seem that CEA has tried merely to push AI safety as the only thing to work on. We don’t think that, and that wasn’t our intention. Obviously, poorly realized intentions are still a problem, but I want to reassure people about CEA’s approach to these issues.
First, re there not being enough discussion of portfolios/comparative advantage, this is mentioned in two of the articles (“Prospecting for Gold” and “What Does (and Doesn’t) AI Mean for Effective Altruism?”). However, I think that we could have emphasised this more, and I will see if it’s possible to include a full article on coordination and comparative advantage.
Second, I’d like to apologise for the way the animal and global health articles came across. Those articles were commissioned at the same time as the long-term future article, and they share a common structure: What’s the case for this cause? What are some common concerns about that cause? Why might you choose not to support this cause? The intention was to show how many assumptions underlie a decision to focus on any cause, and to map out some of the debate between the different cause areas, rather than to illicitly push the long-term future. It looks like this didn’t come across, sorry. We didn’t initially commission sub-cause profiles on government, AI and biosecurity, which explains why those more specific articles follow a different structure (mostly talks given at EA Global).
Third, I want to explain some of the reasoning behind including several articles on AI. AI risk is a more unusual area, which is more susceptible to misinterpretation than global health or animal welfare. Partly for this reason, we thought that it was sensible to include several articles on this topic, with the intention that this would provide more needed background and convey more of the nuance of the idea. I will talk with some of the commenters above to discuss if it makes sense to do some sort of merge so that AI dominates the contents page less.
What about the possibility the Centre for Effective Altruism represents the community by editing the EA Handbook to reflect what the community values in spite of what the CEA concludes, the CEA excludes evaluations from the EA Handbook from which it currently diverges from the community, and it’s still called the ‘EA Handbook’ instead of ‘CEA’s Guide to EA?’ Obviously this wouldn’t carry EA forward with what the CEA thinks is maximum fidelity, but it’s clear many think the CEA is trying to spread the EA message with infidelity, while acting as though they’re the only actor in the movement others can trust to carry that message. It looks not only hypocritical but undermines faith in the CEA.
Altering the handbook so it’s more of a compromise between multiple actors in EA will redeem the reputation of the CEA. Without that, the CEA can’t carry EA forward with the fidelity at all, because the rest of the movement wouldn’t cooperate with them. In the meantime, the CEA and everyone else can hammer out what we think is the most good here on the EA Forum. If broader conclusions are drawn which line up with the CEA’s evaluation based on a consensus the CEA had behind their perspective the best arguments, that can be included in the next edition of the EA Handbook. Again, from the CEA’s perspective, that might seem like deliberately compromising the fidelity of EA in the short-term to appease others. But again, from the perspective of the CEA’s current critics, why they’re criticizing the 2nd edition of the EA Handbook is because they perceive themselves as protecting the fidelity of EA from the Centre for Effective Altruism. This could solve other contentious issues in EA, such as consideration of both s-risks and x-risks from AI. The EA Handbook could be published as close to identically as possible in multiple languages, which would prevent the CEA from selling EA one way in English, the EAF selling it another way in German, and creating more trust issues which would down the road just become sources of conflict, not unlike the criticism the EA Handbook, 2nd edition, is receiving now. Ultimately, this would be the CEA making a relatively short-term compromise to ensure the long-term fidelity of EA by demonstrating themselves as delegate and representative agency the EA community can still have confidence in.
Thanks for the comments Evan. First, I want to apologize for not seeking broader consultation earlier. This was clearly a mistake.
My plan now is to do as you suggest: talk to other actors in EA and get their feedback on what to include etc. Obviously any compromise is going to leave some unhappy—different groups do just favour different presentations of EA, so it seems unlikely to me that we will get a fully independent presentation that will please everyone. I also worry that democracy is not well suited to editorial decisions, and that the “electorate” of EA is ill-defined. If the full compromise approach fails, I think it would be best to release a CEA-branded resource which incorporates most of the feedback above. This option also seems to me to be cooperative, and to avoid harm to the fidelity of EA’s message, but I might be missing something.
Thanks for responding Max. I agree consulting some key actors but not going through a democratic makes sense. I appreciate you being able to respond to and incorporate all the feedback you’re receiving so quickly.