To the extent that this is saying that you expect Lightcone to pay fewer (unjustified) costs than most EA orgs from people publishing criticisms without pinging us first, I think that’s probably true. Those costs are obviously not zero and it’s not obvious to me that the “correct” policy for any given individual who might want to publish such a criticism is to skip that step by default, if they’re optimizing for the propagation of true information.
The reason most people should at least consider skipping that step is because, as Jeff points out, this expectation is a non-trivial cost that will sometimes cause the publication to not happen. However, if you’re a person for whom the cost of that kind of check-in is low, there may often be an intersection above zero between the curves of your cost and the improved output you’d get from the check-in.
I’m not saying it won’t improve someone’s post to get direct feedback from us, and I’m not saying it might not end up reducing some amount of effort from someone on the lightcone team to respond to things people are wrong about, but my current model is that for people to have justified belief in their model of the work that an org does, they should believe they would have heard negative info about us if it exists, and so I ought to encourage people to be openly severely critical and push back against demands to not write their criticism for a pretty large swathe of possible reasons.
Concretely, public criticism of CFAR and MIRI has made me feel more confident in my model of how bad things have been in those places (and how bad it hasn’t). Even if the criticism itself was costly to respond to, the cost was just worth it in terms of people being able to come to accurate beliefs about those places.
I wasn’t referring specifically to the OP when I wrote that, I meant I ought to pushback against a pretty wide swath of possible arguments and pressures against publishing criticism. Nonetheless I want to answer your question.
My answer is more “yes” than “no”. If someone is to publish a critique of an EA org and hasn’t shown it to the org, people can say “But you didn’t check it with the org, which we agreed is a norm around here, so you don’t seem to be willing to play by the rules, and I now suspect the rest of this post is in bad faith.” Yet I think it’s a pretty bad norm in many important instances. If the writer feels personally harmed or tricked by the org, they often will feel that their relationship with the org is rocky, and may choose to say nothing rather than talk to the org. I am reminded of a time at university where I felt so scrutinized and disliked by my professors that after completing my assigned problem set, I couldn’t bring myself to hand it in and face them directly. Somewhat more strongly, I think requiring people to talk to whoever they’re saying negative things about would have made it much more costly for the authors of both of the links I linked to above, and I think it’s better that they published than increase the chance by ~40% that they wouldn’t publish at all, or a year later. If I consider a former FTX/Alameda employee who felt gaslit and tricked by the management in those places (pre the crash), if they had chosen to write about it, I think they would have felt actively (and correctly) scared about privately reaching out to the organization first for fear of retaliation, and again many people would prefer to say nothing rather than have that direct interaction where they say “I’m going to say negative things about you publicly”.
(Also Zoe Curzi and Leverage. Really there’s a lot of examples.)
My main issue with the norm you’re proposing is that often people are (a) scared about retaliation, and even worse (b) scared to mention on the meta-level that they’re scared, again for fear of retaliation. As such, additional norms required for interacting directly with the people of whom you’re scared, seem to super strongly punish a substantial class of very important criticism from being published at all.
(Also Zoe Curzi and Leverage. Really there’s a lot of examples.)
Also examples on the other side, I would note. Without a healthy skepticism of anonymous or other kinds of reduced-accountability reports, one would’ve been lead around by the nose by Ziz’s attempts.
I ought to pushback against a pretty wide swath of possible arguments and pressures against publishing criticism
If someone is to publish a critique of an EA org and hasn’t shown it to the org, people can say “But you didn’t check it with the org, which we agreed is a norm around here, …”
Aha. Now it seems to me that my reading of the OP and my reaction, as well as others’s reading of my comments, have both followed this pattern:
P1 implicitly proposes to call on some social machinery in some way (in jefftk’s case, the machinery of norm setting; in my case, the machinery of group-epistemology)
P2 objects to a wide swath of proposals to call on that machinery (you, me, others pushing back on this norm setting; jefftk and others push back against trusting group epistemology)
P1 is confused about the response to some other proposal, or the imputation of claims not made
In both cases I think that the most salient thing should be: this social machinery is broken. The norm setting/enforcing machine is broken, and the group epistemology machine is broken.
In cases where you’re worried about bad behavior by an org or have had a bad experience with them and don’t want to interact with them (including the examples you described above) I agree it’s fine to go ahead without sending it to them.
On the other hand, I think this is only rarely the case for critical posts? The larger category, what this doesn’t apply, is what I was trying to address here. I should edit the post to include this, though I need to think about the wording and don’t have time to make the change right now.
Right. I suspect we still have some disagreement but happy to leave it here.
(To briefly leave pointer, but with no expectation Jeff for you to respond to it: I think this sort of dynamic extends further into lots of other criticism, where even if your criticism isn’t about bad behavior you’re still pretty unsure how they respond to criticism and whether they’ll respond well, and it can be very stressful to engage directly yet still pro-social to publish criticism.)
For the record I don’t think anyone needs to check with Lightcone before criticizing any of our work.
To the extent that this is saying that you expect Lightcone to pay fewer (unjustified) costs than most EA orgs from people publishing criticisms without pinging us first, I think that’s probably true. Those costs are obviously not zero and it’s not obvious to me that the “correct” policy for any given individual who might want to publish such a criticism is to skip that step by default, if they’re optimizing for the propagation of true information.
The reason most people should at least consider skipping that step is because, as Jeff points out, this expectation is a non-trivial cost that will sometimes cause the publication to not happen. However, if you’re a person for whom the cost of that kind of check-in is low, there may often be an intersection above zero between the curves of your cost and the improved output you’d get from the check-in.
I’m not saying it won’t improve someone’s post to get direct feedback from us, and I’m not saying it might not end up reducing some amount of effort from someone on the lightcone team to respond to things people are wrong about, but my current model is that for people to have justified belief in their model of the work that an org does, they should believe they would have heard negative info about us if it exists, and so I ought to encourage people to be openly severely critical and push back against demands to not write their criticism for a pretty large swathe of possible reasons.
Concretely, public criticism of CFAR and MIRI has made me feel more confident in my model of how bad things have been in those places (and how bad it hasn’t). Even if the criticism itself was costly to respond to, the cost was just worth it in terms of people being able to come to accurate beliefs about those places.
Do you read my post as a demand that people not write their criticism?
I wasn’t referring specifically to the OP when I wrote that, I meant I ought to pushback against a pretty wide swath of possible arguments and pressures against publishing criticism. Nonetheless I want to answer your question.
My answer is more “yes” than “no”. If someone is to publish a critique of an EA org and hasn’t shown it to the org, people can say “But you didn’t check it with the org, which we agreed is a norm around here, so you don’t seem to be willing to play by the rules, and I now suspect the rest of this post is in bad faith.” Yet I think it’s a pretty bad norm in many important instances. If the writer feels personally harmed or tricked by the org, they often will feel that their relationship with the org is rocky, and may choose to say nothing rather than talk to the org. I am reminded of a time at university where I felt so scrutinized and disliked by my professors that after completing my assigned problem set, I couldn’t bring myself to hand it in and face them directly. Somewhat more strongly, I think requiring people to talk to whoever they’re saying negative things about would have made it much more costly for the authors of both of the links I linked to above, and I think it’s better that they published than increase the chance by ~40% that they wouldn’t publish at all, or a year later. If I consider a former FTX/Alameda employee who felt gaslit and tricked by the management in those places (pre the crash), if they had chosen to write about it, I think they would have felt actively (and correctly) scared about privately reaching out to the organization first for fear of retaliation, and again many people would prefer to say nothing rather than have that direct interaction where they say “I’m going to say negative things about you publicly”.
(Also Zoe Curzi and Leverage. Really there’s a lot of examples.)
My main issue with the norm you’re proposing is that often people are (a) scared about retaliation, and even worse (b) scared to mention on the meta-level that they’re scared, again for fear of retaliation. As such, additional norms required for interacting directly with the people of whom you’re scared, seem to super strongly punish a substantial class of very important criticism from being published at all.
Also examples on the other side, I would note. Without a healthy skepticism of anonymous or other kinds of reduced-accountability reports, one would’ve been lead around by the nose by Ziz’s attempts.
Aha. Now it seems to me that my reading of the OP and my reaction, as well as others’s reading of my comments, have both followed this pattern:
P1 implicitly proposes to call on some social machinery in some way (in jefftk’s case, the machinery of norm setting; in my case, the machinery of group-epistemology)
P2 objects to a wide swath of proposals to call on that machinery (you, me, others pushing back on this norm setting; jefftk and others push back against trusting group epistemology)
P1 is confused about the response to some other proposal, or the imputation of claims not made
In both cases I think that the most salient thing should be: this social machinery is broken. The norm setting/enforcing machine is broken, and the group epistemology machine is broken.
In cases where you’re worried about bad behavior by an org or have had a bad experience with them and don’t want to interact with them (including the examples you described above) I agree it’s fine to go ahead without sending it to them.
On the other hand, I think this is only rarely the case for critical posts? The larger category, what this doesn’t apply, is what I was trying to address here. I should edit the post to include this, though I need to think about the wording and don’t have time to make the change right now.
Right. I suspect we still have some disagreement but happy to leave it here.
(To briefly leave pointer, but with no expectation Jeff for you to respond to it: I think this sort of dynamic extends further into lots of other criticism, where even if your criticism isn’t about bad behavior you’re still pretty unsure how they respond to criticism and whether they’ll respond well, and it can be very stressful to engage directly yet still pro-social to publish criticism.)
Edited to add something covering this, though I suspect it doesn’t go as far as you’d prefer?
(Also curious what you think of Ray’s argument)
I actually think your caveat helps a lot.