It’s misleading to frame the argument as “them claiming things were inaccurate was appropriate reason to delay publication.” The appropriate reason to delay publication was their evident willingness to compile specific counter-evidence within a week. Of course subjects of hostile articles will always claim inaccuracies, but it matters whether they can credibly claim ability to provide contrary information. “Very last-minute screenshots” simply should not be a thing when working on an investigative piece of this magnitude—if you’re doing investigative work, you have a duty to do it right, not call it short based on the “time [you] had budgeted” and publish whatever you have.
Here’s my standard: epistemic disclaimers do not matter much when it comes to articles impugning reputations. What matters is presenting all available information accurately to the best of your ability. Your claim that the alternative was not a delay simply does not hold water: whether the alternative was a delay or no publication at all was fully within your control. There’s no reason to suspect any retaliation would be greater without publication than with publication; your time budget is nobody’s concern but your own. If the post contained a single meaningful falsehood at publication that could have been prevented by reviewing the information the subject of the article was actively preparing for you, publication at the chosen time was unreasonable no matter how many disclaimers Ben included.
If the post contained a single meaningful falsehood at publication that could have been prevented by reviewing the information the subject of the article was actively preparing for you, publication at the chosen time was unreasonable no matter how many disclaimers Ben included.
This is a universal statements that’s clearly inaccurate. The relevance of the falsehoods to the central case really matters (if it turns out that a source got a number off by a single irrelevant digit, or it go the name of a city wrong that could have just been omitted, etc.). Any article of this size will have some inaccuracies in them. I agree that there should still be a pretty harsh tradeoff towards accuracy, though in situations like this with very credible evidence that information was being heavily suppressed from being shared (which I still believe), it is also a high priority to get anything out.
It would be terrible if someone had evidence of fraud at FTX, but didn’t have the time to publish it because in order to make their case they had to spend thousands of hours getting each individual detail of their story exactly right.
Separately, it is totally normal, and I don’t see an alternative, that sometimes the thing you do is directly and accurately report what a source has told you. You don’t endorse what the source said, but you just directly state it.
Journalists aren’t responsible for the accuracy of every single thing their sources say, they are responsible for accurately citing reporting what their sources say. If you have another source disputing your first source, you accurately summarize that contradicting source too. Sometimes people say wrong things. Sometimes the fact that they say wrong things is even materially relevant to the story.
It clearly must be possible to write an article that includes the sentence “this source says X” even if X is wrong, or if you think there is only a 50% chance that X is true.
“Meaningful” covers cases like the ones you mentioned. I stand by my words.
Journalists are responsible, to the best of my understanding, for the accuracy of every single thing they say, which includes the things their sources say. If a source says something a journalist knows to be false and the journalist reports that claim, knowing it to be false, they are not fulfilling their duty. As far as I can observe, this aligns with the legal standard (as I discuss here) as well as the ethical standard.
When you amplify someone’s claims, you take responsibility for those claims. When you amplify false claims where contradictory evidence is available to you and you decline to investigate that contradictory evidence, you take responsibility for that.
If someone had evidence of fraud at FTX, they should have published specifically the limited set of evidence they were confident in and could independently verify. If they lacked the time to build a more cohesive, complete story, they should have found someone who had that time.
People live and die on their reputations, and spreading falsehoods that damage someone’s reputation is and should be seen as more than just a minor faux pas. I understand the environment that makes EAs want to overcorrect on this right now, but due diligence is not optional when whistleblowing.
I don’t think a norm that if you quote someone or try to comment on something someone said, you have to take full responsibility for the accuracy of their statements is a recipe for good public communication.
When I write on LessWrong, or the EA Forum, or when I talk to my friends in Slack and large chat rooms, I would absolutely not hold people responsible for the content of things they are commenting on. Indeed, I think it would be impossible for me to have almost any of the conversations I routinely have, if I did try that.
This kind of relationship to publication seems to me to have enormous chilling effects on information propagation and communication, and I don’t really see its merits.
Yes, of course I will sometimes cite and refer sources that I can’t fully verify as reliable. In those cases I will state that I don’t think they are fully reliable. This seems so much better than being unable to express my epistemic state at all.
If I believe something, I try to give a detailed account of what evidence has convinced me of that belief. This of course will often include sources that aren’t fully reliable (as practically no source is). And the end state of my belief is often one of uncertainty, and it seems crucial to be able to express this uncertainty.
I am planning to continue going around and share many concerns I have about people, even if I can’t definitely prove them beyond the shadow of a doubt. I think it’s OK for other people to operate on a different philosophy, but I really don’t feel very beholden or resonate much with the standard you outline here.
It’s a category error, in my estimation, to lump casual commentary in with published articles. You and Ben published a thorough, detailed article that mixed hundreds of hours of investigative reporting and opinion with a very specific goal: Destroy the reputation of a group within your community. It is currently the sixth-most popular post of the year, which I believe is a significant decline from its peak, and had massive impacts on your community and subculture.
That is something altogether different from an offhand comment deep in a thread. It is a massive undertaking with potentially massive ramifications in the lives of other people and as such entails a massive duty to those people. What are the merits of this kind of relationship to publication—a kind, I want to emphasize, that every journalistic institution in the United States claims? Very simple: if you have that kind of relationship to publication, you do not cause damage to others via propagation of false rumors about them.
Journalists don’t have a very good reputation right now, and not without reason, but a responsibility for accuracy is core to the first, second, and third principles in their code of ethics. “They were simply too loyal to truth, and too cautious about publishing things that later turned out to be false” is not a criticism I have ever heard leveled at a journalistic institution.
The standard is not “beyond a shadow of a doubt,” but a minimum standard is absolutely “wait to review immediately forthcoming contradictory evidence before publishing falsehood.”
I admire the first-principles thinking of rationalists and effective altruists. In many ways, it resonates with me. But the standards of common law and of broader ethics are written in blood, and outright dismissing those standards due to confidence that your own first-principles thinking is superior not only leaves you open to liability any time your actions clash with those standards, it leaves you in a position to cause much of the same damage those standards were written to prevent.
You and Ben published a thorough, detailed article that mixed hundreds of hours of investigative reporting and opinion with a very specific goal: Destroy the reputation of a group within your community. It is currently the sixth-most popular post of the year, which I believe is a significant decline from its peak, and had massive impacts on your community and subculture.
Just for the record, this is inaccurate in a bunch of different ways.
Our goal was not to destroy the reputation of that group within our community. Our primary goal was to investigate some bad things we heard about, with honestly my primary motivation being to figure out ways to somehow allow the very extensive rumors and warnings from many dozens of people that I’ve heard about in the case of FTX, to translate into further investigation and scrutiny.
In the case of FTX, there were similarly an enormous number of rumors and flags that people had raised, but nobody felt comfortable publishing those, which played a major role in the damage that FTX was able to cause. At least my primary goal in allocating resources to this was to try very concretely to apply those lessons to the Nonlinear case and see whether it’s possible to somehow get information from the rumor level into a more fact-checkable and verifiable level.
It is currently the sixth-most popular post of the year, which I believe is a significant decline from its peak
I think this is also false. The post lost about 20 karma in the last few days, not really changing its relative ranking at all.
At the time of publication, I struggle to differentiate between “destroy the reputation” and “allow the very extensive rumors and warnings from many dozens of people … to translate into further investigation and scrutiny [comparable to the case of FTX, the biggest scandal in your community]”. I respect that you perceive a difference but the one seems like an expansion of the other. Raising serious red flags about an organization has straightforward and predictable reputational effects.
I’ll strikethrough the other claim—I must have misinterpreted some of what I remembered reading about it.
At the time of publication, I struggle to differentiate between “destroy the reputation” and “allow the very extensive rumors and warnings from many dozens of people … to translate into further investigation and scrutiny [comparable to the case of FTX, the biggest scandal in your community]”. I respect that you perceive a difference but the one seems like an expansion of the other. Raising serious red flags about an organization has straightforward and predictable reputational effects.
Hmm, I feel like what I am trying to say here is not that weird. My current model is that if you want to be able to catch things like FTX early, before they explode as badly as they did and cause enormous harms, you need some better ways of propagating information than we had at the time. Information about sketchiness and wrongdoing was present, but definitive proof of fraud was not present.
Of course most of the time when the community notices flags like this it will not turn out to be as big as we found out FTX was after it exploded. If I had definitive knowledge that FTX was defrauding billions of customer deposits I would have taken very different actions. But there must be some way to propagate my concerns about FTX pre-collapse which maybe could have reduced the size of the fallout, or sparked an investigation that uncovered the full extend of the fraud. If you have better ways for doing that than the kind of investigation we tried to do here, I would be very interested in hearing them.
It’s not weird, it’s just that there might be a disconnect in tone. “Destroy reputation” has negative connotations; “raise red flags” has responsible connotations. It can be correct to damage someone’s reputation if they do things incompatible with that reputation, and raising red flags has a way of doing that.
I think I’ve been clear and consistent in terms of “better ways.” When people raise concerns, they should be investigated. When you’re looking to publish those concerns, you should make reasonable efforts to verify their specifics, which matter a great deal and cannot be handwaved away. If you are reporting secondhand and not from direct experience, you should either receive primary source documents that let you see evidence directly or check to see whether the accused can provide contrary evidence on every claim in dispute.
In this case, every claim that was important enough to be included in the document was important enough to verify with either direct primary evidence or against Nonlinear’s word and evidence. There are multiple unambiguous instances in the article where you relayed false information that was contradicted by evidence Nonlinear was volunteering to make available to you. You excuse those by claiming that pressure meant you had a duty to release. It did not. Pressure is the correct response to imminent publication of falsehoods and is only inappropriate against truth. The greater the pressure, the more it matters whether you wind up presenting the truth on every particular in your report.
In numbered format:
If Alice or Chloe had wanted to share their direct personal experience, they could have done so without using you as an intermediary, taking personal responsibility for the truth or falsehood of their claims.
Once you elected to be intermediary, you should have relied on primary source documents and gathered as much of it as possible to support or contradict various claims. You should not have turned down a request to send you detailed primary source documents, nor should you have neglected to update your story on receiving new ones even two hours before publication.
In the absence of primary source documents for or against a given claim, you would be free to share it as a claim from an involved party reliant on their word.
The only appropriate time to publish is when the report is complete—when every claim you want to publish has been vetted and is good to go. External and internal pressure, even threats of lawsuits, matter massively less than whether you have given a complete and accurate accounting of what you were aiming to investigate.
I think you’ve both raised good points. Way upthread @Habryka said “I don’t see a super principled argument for giving two weeks instead of one week”, but if I were unfairly accused I’d certainly want a full two weeks! So Kat’s request for a full week to gather evidence seems reasonable [ed: under the principle of due process], and I don’t see what sort of opportunities would’ve existed for retribution from K&E in the two-week case that didn’t exist in the one-week case.
However, when I read Ben’s post (like TW, I did this “fresh” about two days ago; I didn’t see Ben’s post until Kat’s post was up) it sounds like there was more evidence behind it than he specifically detailed (e.g. “I talked to many people who interacted with Emerson and Kat who had many active ethical concerns about them and strongly negative opinions”). Given this, plus concerning aspects of Kat’s response, I think Ben’s post is probably broadly accurate―perhaps overbiased against NL based on the evidence I’ve seen, but perhaps that’s compensated by evidence I haven’t seen, that was only alluded to.
(Edit: but it also seems like the wording of Ben’s piece would’ve softened if they’d waited a bit longer, so… basically I lean more toward TW’s position. But also, I don’t expect the wording to have softened that much. This is all so damn nuanced! Also, I actually think even a partial softening of Ben’s post would’ve been important and might have materially changed Kat’s response and increased community cohesion. K&E likely have personality flaws, but are also likely EAs and rationalists at heart. I respect that, and I respect the apparently substantial funds they put into trying to do good, and so it seems like it would’ve been worth spending more time to get Ben’s initial post right. I’m sad about this situation, I guess because I feel that both Ben and Kat’s posts were worded in somewhat unfair ways, and I’m unconvinced that quite so much acrimony was necessary.)
It’s misleading to frame the argument as “them claiming things were inaccurate was appropriate reason to delay publication.” The appropriate reason to delay publication was their evident willingness to compile specific counter-evidence within a week. Of course subjects of hostile articles will always claim inaccuracies, but it matters whether they can credibly claim ability to provide contrary information. “Very last-minute screenshots” simply should not be a thing when working on an investigative piece of this magnitude—if you’re doing investigative work, you have a duty to do it right, not call it short based on the “time [you] had budgeted” and publish whatever you have.
Here’s my standard: epistemic disclaimers do not matter much when it comes to articles impugning reputations. What matters is presenting all available information accurately to the best of your ability. Your claim that the alternative was not a delay simply does not hold water: whether the alternative was a delay or no publication at all was fully within your control. There’s no reason to suspect any retaliation would be greater without publication than with publication; your time budget is nobody’s concern but your own. If the post contained a single meaningful falsehood at publication that could have been prevented by reviewing the information the subject of the article was actively preparing for you, publication at the chosen time was unreasonable no matter how many disclaimers Ben included.
This is a universal statements that’s clearly inaccurate. The relevance of the falsehoods to the central case really matters (if it turns out that a source got a number off by a single irrelevant digit, or it go the name of a city wrong that could have just been omitted, etc.). Any article of this size will have some inaccuracies in them. I agree that there should still be a pretty harsh tradeoff towards accuracy, though in situations like this with very credible evidence that information was being heavily suppressed from being shared (which I still believe), it is also a high priority to get anything out.
It would be terrible if someone had evidence of fraud at FTX, but didn’t have the time to publish it because in order to make their case they had to spend thousands of hours getting each individual detail of their story exactly right.
Separately, it is totally normal, and I don’t see an alternative, that sometimes the thing you do is directly and accurately report what a source has told you. You don’t endorse what the source said, but you just directly state it.
Journalists aren’t responsible for the accuracy of every single thing their sources say, they are responsible for accurately citing reporting what their sources say. If you have another source disputing your first source, you accurately summarize that contradicting source too. Sometimes people say wrong things. Sometimes the fact that they say wrong things is even materially relevant to the story.
It clearly must be possible to write an article that includes the sentence “this source says X” even if X is wrong, or if you think there is only a 50% chance that X is true.
“Meaningful” covers cases like the ones you mentioned. I stand by my words.
Journalists are responsible, to the best of my understanding, for the accuracy of every single thing they say, which includes the things their sources say. If a source says something a journalist knows to be false and the journalist reports that claim, knowing it to be false, they are not fulfilling their duty. As far as I can observe, this aligns with the legal standard (as I discuss here) as well as the ethical standard.
When you amplify someone’s claims, you take responsibility for those claims. When you amplify false claims where contradictory evidence is available to you and you decline to investigate that contradictory evidence, you take responsibility for that.
If someone had evidence of fraud at FTX, they should have published specifically the limited set of evidence they were confident in and could independently verify. If they lacked the time to build a more cohesive, complete story, they should have found someone who had that time.
People live and die on their reputations, and spreading falsehoods that damage someone’s reputation is and should be seen as more than just a minor faux pas. I understand the environment that makes EAs want to overcorrect on this right now, but due diligence is not optional when whistleblowing.
I don’t think a norm that if you quote someone or try to comment on something someone said, you have to take full responsibility for the accuracy of their statements is a recipe for good public communication.
When I write on LessWrong, or the EA Forum, or when I talk to my friends in Slack and large chat rooms, I would absolutely not hold people responsible for the content of things they are commenting on. Indeed, I think it would be impossible for me to have almost any of the conversations I routinely have, if I did try that.
This kind of relationship to publication seems to me to have enormous chilling effects on information propagation and communication, and I don’t really see its merits.
Yes, of course I will sometimes cite and refer sources that I can’t fully verify as reliable. In those cases I will state that I don’t think they are fully reliable. This seems so much better than being unable to express my epistemic state at all.
If I believe something, I try to give a detailed account of what evidence has convinced me of that belief. This of course will often include sources that aren’t fully reliable (as practically no source is). And the end state of my belief is often one of uncertainty, and it seems crucial to be able to express this uncertainty.
I am planning to continue going around and share many concerns I have about people, even if I can’t definitely prove them beyond the shadow of a doubt. I think it’s OK for other people to operate on a different philosophy, but I really don’t feel very beholden or resonate much with the standard you outline here.
It’s a category error, in my estimation, to lump casual commentary in with published articles. You and Ben published a thorough, detailed article that mixed hundreds of hours of investigative reporting and opinion with a very specific goal: Destroy the reputation of a group within your community. It is currently the sixth-most popular post of the year
, which I believe is a significant decline from its peak,and had massive impacts on your community and subculture.That is something altogether different from an offhand comment deep in a thread. It is a massive undertaking with potentially massive ramifications in the lives of other people and as such entails a massive duty to those people. What are the merits of this kind of relationship to publication—a kind, I want to emphasize, that every journalistic institution in the United States claims? Very simple: if you have that kind of relationship to publication, you do not cause damage to others via propagation of false rumors about them.
Journalists don’t have a very good reputation right now, and not without reason, but a responsibility for accuracy is core to the first, second, and third principles in their code of ethics. “They were simply too loyal to truth, and too cautious about publishing things that later turned out to be false” is not a criticism I have ever heard leveled at a journalistic institution.
The standard is not “beyond a shadow of a doubt,” but a minimum standard is absolutely “wait to review immediately forthcoming contradictory evidence before publishing falsehood.”
I admire the first-principles thinking of rationalists and effective altruists. In many ways, it resonates with me. But the standards of common law and of broader ethics are written in blood, and outright dismissing those standards due to confidence that your own first-principles thinking is superior not only leaves you open to liability any time your actions clash with those standards, it leaves you in a position to cause much of the same damage those standards were written to prevent.
Just for the record, this is inaccurate in a bunch of different ways.
Our goal was not to destroy the reputation of that group within our community. Our primary goal was to investigate some bad things we heard about, with honestly my primary motivation being to figure out ways to somehow allow the very extensive rumors and warnings from many dozens of people that I’ve heard about in the case of FTX, to translate into further investigation and scrutiny.
In the case of FTX, there were similarly an enormous number of rumors and flags that people had raised, but nobody felt comfortable publishing those, which played a major role in the damage that FTX was able to cause. At least my primary goal in allocating resources to this was to try very concretely to apply those lessons to the Nonlinear case and see whether it’s possible to somehow get information from the rumor level into a more fact-checkable and verifiable level.
I think this is also false. The post lost about 20 karma in the last few days, not really changing its relative ranking at all.
At the time of publication, I struggle to differentiate between “destroy the reputation” and “allow the very extensive rumors and warnings from many dozens of people … to translate into further investigation and scrutiny [comparable to the case of FTX, the biggest scandal in your community]”. I respect that you perceive a difference but the one seems like an expansion of the other. Raising serious red flags about an organization has straightforward and predictable reputational effects.
I’ll strikethrough the other claim—I must have misinterpreted some of what I remembered reading about it.
Hmm, I feel like what I am trying to say here is not that weird. My current model is that if you want to be able to catch things like FTX early, before they explode as badly as they did and cause enormous harms, you need some better ways of propagating information than we had at the time. Information about sketchiness and wrongdoing was present, but definitive proof of fraud was not present.
Of course most of the time when the community notices flags like this it will not turn out to be as big as we found out FTX was after it exploded. If I had definitive knowledge that FTX was defrauding billions of customer deposits I would have taken very different actions. But there must be some way to propagate my concerns about FTX pre-collapse which maybe could have reduced the size of the fallout, or sparked an investigation that uncovered the full extend of the fraud. If you have better ways for doing that than the kind of investigation we tried to do here, I would be very interested in hearing them.
It’s not weird, it’s just that there might be a disconnect in tone. “Destroy reputation” has negative connotations; “raise red flags” has responsible connotations. It can be correct to damage someone’s reputation if they do things incompatible with that reputation, and raising red flags has a way of doing that.
I think I’ve been clear and consistent in terms of “better ways.” When people raise concerns, they should be investigated. When you’re looking to publish those concerns, you should make reasonable efforts to verify their specifics, which matter a great deal and cannot be handwaved away. If you are reporting secondhand and not from direct experience, you should either receive primary source documents that let you see evidence directly or check to see whether the accused can provide contrary evidence on every claim in dispute.
In this case, every claim that was important enough to be included in the document was important enough to verify with either direct primary evidence or against Nonlinear’s word and evidence. There are multiple unambiguous instances in the article where you relayed false information that was contradicted by evidence Nonlinear was volunteering to make available to you. You excuse those by claiming that pressure meant you had a duty to release. It did not. Pressure is the correct response to imminent publication of falsehoods and is only inappropriate against truth. The greater the pressure, the more it matters whether you wind up presenting the truth on every particular in your report.
In numbered format:
If Alice or Chloe had wanted to share their direct personal experience, they could have done so without using you as an intermediary, taking personal responsibility for the truth or falsehood of their claims.
Once you elected to be intermediary, you should have relied on primary source documents and gathered as much of it as possible to support or contradict various claims. You should not have turned down a request to send you detailed primary source documents, nor should you have neglected to update your story on receiving new ones even two hours before publication.
In the absence of primary source documents for or against a given claim, you would be free to share it as a claim from an involved party reliant on their word.
The only appropriate time to publish is when the report is complete—when every claim you want to publish has been vetted and is good to go. External and internal pressure, even threats of lawsuits, matter massively less than whether you have given a complete and accurate accounting of what you were aiming to investigate.
I think you’ve both raised good points. Way upthread @Habryka said “I don’t see a super principled argument for giving two weeks instead of one week”, but if I were unfairly accused I’d certainly want a full two weeks! So Kat’s request for a full week to gather evidence seems reasonable [ed: under the principle of due process], and I don’t see what sort of opportunities would’ve existed for retribution from K&E in the two-week case that didn’t exist in the one-week case.
However, when I read Ben’s post (like TW, I did this “fresh” about two days ago; I didn’t see Ben’s post until Kat’s post was up) it sounds like there was more evidence behind it than he specifically detailed (e.g. “I talked to many people who interacted with Emerson and Kat who had many active ethical concerns about them and strongly negative opinions”). Given this, plus concerning aspects of Kat’s response, I think Ben’s post is probably broadly accurate―perhaps overbiased against NL based on the evidence I’ve seen, but perhaps that’s compensated by evidence I haven’t seen, that was only alluded to.
(Edit: but it also seems like the wording of Ben’s piece would’ve softened if they’d waited a bit longer, so… basically I lean more toward TW’s position. But also, I don’t expect the wording to have softened that much. This is all so damn nuanced! Also, I actually think even a partial softening of Ben’s post would’ve been important and might have materially changed Kat’s response and increased community cohesion. K&E likely have personality flaws, but are also likely EAs and rationalists at heart. I respect that, and I respect the apparently substantial funds they put into trying to do good, and so it seems like it would’ve been worth spending more time to get Ben’s initial post right. I’m sad about this situation, I guess because I feel that both Ben and Kat’s posts were worded in somewhat unfair ways, and I’m unconvinced that quite so much acrimony was necessary.)