Linch, surprised you felt like titotal wasn’t reading your comment properly, since I feel like they make a version of the basically right argument here which is around deterrence and the benefits of public knowledge of wrongdoing outside the specific case. Any sort of investigatory/punitive process (e.g. in most legal contexts) will often have resources devoted to it that are very significant compared to the actual potential wrongdoing being interrogated. But having a system that reliably identifies wrongdoing is quite valuable (and even a patchwork system is probably also quite valuable). Plus there are a whole bunch of diffuse positive externalities to information (e.g. not requiring each actor in the system to spend the effort making a private judgment that has a decent chance of being wrong).
I think the broader problem with your argument here is it’s an example of consequentialism struggling to deal with collective action problems/the value of institutions. The idea that all acts can be cashed out into utility (i.e. “world is burning” above) struggles to engage with cases where broader institutions are necessary for an ecosystem to function. To use an example from outside this case, if one evaluates public statements on their individual utility (rather than descriptive accuracy), it can stymie free inquiry and lead to poorer decision-making. (Not saying this can never be accounted for through a consequentialist or primarily consequentialist theory but I think it’s a persistent and difficult problem).
I think “you didn’t seem to read my comment, which frustrates me” is a better thing to say to someone than “are you a native english speaker?” since it seems to get at the problem more directly and isn’t exclusionary to non-native speakers (which is rude, even if that’s not the intention). I also think the instant case should give pause about the way you’re attempting to deal with bad faith critics, since labeling a critic mentally as poorly comprehending or in bad faith can be a subconscious crutch to miss the thrust of their argument.
On the object-level, I think the argument has to route around something like “1 public investigation is worth 3-10 private cases not being addressed/punished,” because in the world we live in, resources are patently limited and most people empirically do not go around investigating potential bad actors.
I feel pretty much capped at how much time/emotional energy I can spend on investigations, given that I have competing priorities like a day job and writing snarky EA forum comments. I already kinda feel like “did my time” for things-akin-to-investigations that I initiated or participated in; if the expectation is always a public writeup I might as well curl up in a ball or something. And I suspect I’m more able/willing to take on public flak and private reprisal than most people on this forum. So I expect having norms to make investigations more costly to straightforwardly decrease them happening.
I think an undertone in much of this thread is that people simultaneously expect processes that can “reliably identify wrongdoing” as well as the benefits of public discourse, without being willing to pay the costs associated with such investigations and disclosures. It reminds me of taxpayers who want lower taxes and increased budgets. Overall this feels surprisingly “not EA” to me, in that one of the basic tenets of EA is appreciating the existence of tradeoffs.
I think “you didn’t seem to read my comment, which frustrates me” is a better thing to say to someone than “are you a native english speaker?” since it seems to get at the problem more directly and isn’t exclusionary to non-native speakers
Yeah this is helpful. Though to be clear, I think they did “read” my comment, just didn’t try to understand it.
And I’m sorry to appear to be exclusionary to non-native speakers, though again I want to register that (as you know) I’m not a native speaker myself whereas you and most people reprimanding me appear to be.
Yeah I should have clarified that I knew you’re not a native speaker and understand why that motivates your argument, but the harm of being exclusionary stems in part because not every reader will know that. (Though I think even if every reader did know that you were a non-native speaker, it still does create a negative effect (via this exclusionary channel) albeit a smaller one).
Also I didn’t take your claim to be “investigations should not only take place in cases where their results will be made public.” (Which seems to be the implication of your reply above but maybe I’m misunderstanding). I don’t think “public exposes are useful” implies that you need to necessarily conduct the work needed for a public expose in cases where you suspect wrongdoing.
Should also say as your friend that I recognize it sucks to be criticized especially when it feels like a group pile-on, and I appreciate your making controversial claims even if I don’t agree with them.
but the harm of being exclusionary stems in part because not every reader will know that. (Though I think even if every reader did know that you were a non-native speaker, it still does create a negative effect (via this exclusionary channel) albeit a smaller one).
For the record, I consider being a non-native speaker exculpatory evidence. I’d much rather exclude native speakers with poor[1] reading comprehension than non-native speakers.
Also I didn’t take your claim to be “investigations should not only take place in cases where their results will be made public.” (Which seems to be the implication of your reply above but maybe I’m misunderstanding). I don’t think “public exposes are useful” implies that you need to necessarily conduct the work needed for a public expose in cases where you suspect wrongdoing.
I agree it’s not a necessary result, but I think it has a strong directional effect. At the very least people will feel encouraged to make their results public, a situation where they’re likely to already systematically underestimate the costs (though tbf maybe some of the benefits as well).
Should also say as your friend that I recognize it sucks to be criticized especially when it feels like a group pile-on, and I appreciate your making controversial claims even if I don’t agree with them.
I appreciate the sympathy! I was worried a while ago that I cared more about being liked than being right, so I’m at least glad to get some contrary evidence here.
But the potential to be hurt by bad actors is (Edit: or, arguably, people with less influence and less access to non-public information are actually more likely to get hurt).
Linch, surprised you felt like titotal wasn’t reading your comment properly, since I feel like they make a version of the basically right argument here which is around deterrence and the benefits of public knowledge of wrongdoing outside the specific case. Any sort of investigatory/punitive process (e.g. in most legal contexts) will often have resources devoted to it that are very significant compared to the actual potential wrongdoing being interrogated. But having a system that reliably identifies wrongdoing is quite valuable (and even a patchwork system is probably also quite valuable). Plus there are a whole bunch of diffuse positive externalities to information (e.g. not requiring each actor in the system to spend the effort making a private judgment that has a decent chance of being wrong).
I think the broader problem with your argument here is it’s an example of consequentialism struggling to deal with collective action problems/the value of institutions. The idea that all acts can be cashed out into utility (i.e. “world is burning” above) struggles to engage with cases where broader institutions are necessary for an ecosystem to function. To use an example from outside this case, if one evaluates public statements on their individual utility (rather than descriptive accuracy), it can stymie free inquiry and lead to poorer decision-making. (Not saying this can never be accounted for through a consequentialist or primarily consequentialist theory but I think it’s a persistent and difficult problem).
I think “you didn’t seem to read my comment, which frustrates me” is a better thing to say to someone than “are you a native english speaker?” since it seems to get at the problem more directly and isn’t exclusionary to non-native speakers (which is rude, even if that’s not the intention). I also think the instant case should give pause about the way you’re attempting to deal with bad faith critics, since labeling a critic mentally as poorly comprehending or in bad faith can be a subconscious crutch to miss the thrust of their argument.
On the object-level, I think the argument has to route around something like “1 public investigation is worth 3-10 private cases not being addressed/punished,” because in the world we live in, resources are patently limited and most people empirically do not go around investigating potential bad actors.
I feel pretty much capped at how much time/emotional energy I can spend on investigations, given that I have competing priorities like a day job and writing snarky EA forum comments. I already kinda feel like “did my time” for things-akin-to-investigations that I initiated or participated in; if the expectation is always a public writeup I might as well curl up in a ball or something. And I suspect I’m more able/willing to take on public flak and private reprisal than most people on this forum. So I expect having norms to make investigations more costly to straightforwardly decrease them happening.
I think an undertone in much of this thread is that people simultaneously expect processes that can “reliably identify wrongdoing” as well as the benefits of public discourse, without being willing to pay the costs associated with such investigations and disclosures. It reminds me of taxpayers who want lower taxes and increased budgets. Overall this feels surprisingly “not EA” to me, in that one of the basic tenets of EA is appreciating the existence of tradeoffs.
Yeah this is helpful. Though to be clear, I think they did “read” my comment, just didn’t try to understand it.
And I’m sorry to appear to be exclusionary to non-native speakers, though again I want to register that (as you know) I’m not a native speaker myself whereas you and most people reprimanding me appear to be.
Yeah I should have clarified that I knew you’re not a native speaker and understand why that motivates your argument, but the harm of being exclusionary stems in part because not every reader will know that. (Though I think even if every reader did know that you were a non-native speaker, it still does create a negative effect (via this exclusionary channel) albeit a smaller one).
Also I didn’t take your claim to be “investigations should not only take place in cases where their results will be made public.” (Which seems to be the implication of your reply above but maybe I’m misunderstanding). I don’t think “public exposes are useful” implies that you need to necessarily conduct the work needed for a public expose in cases where you suspect wrongdoing.
Should also say as your friend that I recognize it sucks to be criticized especially when it feels like a group pile-on, and I appreciate your making controversial claims even if I don’t agree with them.
For the record, I consider being a non-native speaker exculpatory evidence. I’d much rather exclude native speakers with poor[1] reading comprehension than non-native speakers.
I agree it’s not a necessary result, but I think it has a strong directional effect. At the very least people will feel encouraged to make their results public, a situation where they’re likely to already systematically underestimate the costs (though tbf maybe some of the benefits as well).
I appreciate the sympathy! I was worried a while ago that I cared more about being liked than being right, so I’m at least glad to get some contrary evidence here.
say worse than GPT-3.5? Or worse than GPT-4, I’m not sure.
As long as a public exposé reaches 3-10X more people, this seems entirely plausible.
Influence in EA is not uniformly distributed, nor is access to information, nor is ability to draw inferences from information.
But the potential to be hurt by bad actors is (Edit: or, arguably, people with less influence and less access to non-public information are actually more likely to get hurt).