I think it’s worth explicitly stating that while I think most in this community are reading “I plan to compare the things that this post says directly against specific claims in mine, and acknowledge anything where I was factually inaccurate” as an example of a positive norm, it reads to me as a strongly negative one.
Specifically: Based on this comment, it sounds like you knew or should have known much of this evidence was available upon the decision to publish your original post, and you went ahead regardless without waiting to review claims that falsify parts of your original post. In other words, you mixed what appear to be materially false claims with true ones in a post aimed explicitly at destroying the reputation of an organization. The correct time to update factual inaccuracies in something so high-stakes is not after-the-fact, it is prior to publication, and the presence of correctable falsehoods in the original, whether or not you subsequently correct them, is a major issue.
Specifically: Based on this comment, it sounds like you knew or should have known much of this evidence was available upon the decision to publish your original post, and you went ahead regardless without waiting to review claims that falsify parts of your original post.
To be clear, as I think I have stated in like dozens of places, at the time the post was written, all evidence that Ben was given made it into the post. There are no glaring omissions, the post accurately captured Ben’s epistemic state at the time.
The only thing happening was that Nonlinear was claiming many of the concrete claims were false, which we did not believe them on (and I still don’t believe them on). Spencer sent us a screenshot about the vegan food stuff 2 hours before publication, which Ben didn’t get around to editing in before the post went live, but that’s all the evidence that I know about that you could somehow argue we had but didn’t include. It is not accurate that Nonlinear sent credible signals of having counterevidence before the post went live (and if someone wants to, I would probably be up for arranging a call with a neutral third party to walk them through the totality of our Nonlinear communications and see whether they find anything that really seems like it had to be included).
(The one other thing that didn’t end up in the post was Alice’s side-business, which we did know about but didn’t consider very important, and Nonlinear didn’t bring it up in the pre-publication conversations we had. Ben is sad about not including it, but it also didn’t seem that material to the case given that it probably never made much money, though me and Ben still think it would have been good to include. However, again, Nonlinear did not poke us about this or share evidence that it was important or was an important omission before publishing)
I responded to your EA forum comments much earlier in the day, but my responses are still stuck in the mod filter, so I might repeat myself a bit here. I’m not part of EA, but I’ve been around the rationalist sphere for a long while and I have more experience than most here with investigative journalism–adjacent work.
You regularly mention hundreds of hours of work, a deadline you had to meet, the inconvenience you would have faced upon delaying, and the chance that you might have chosen not to publish at all had you delayed.
That’s all very well and good, but honestly doesn’t sway the needle for me at all in terms of your duty in the situation. Self-imposed deadlines are not an excuse to avoid hearing out one party to a conflict you’re publishing.
In those hundreds of hours, so far as I can tell, you spent a grand total of three gathering one side of the story—the side of a group Ben’s post was explicitly adversarial towards and explicitly aimed at destroying the reputation of. When they told him they were compiling a detailed list of evidence to respond to some of his claims, you guys flatly refused to investigate further or delay your self-imposed deadline. I cover a story like this almost every week for the podcast I work for, and I put more time into gathering opposing stories than that.
I believe you when you say that (almost) all evidence Ben was given made it into the post. I emphatically disagree that holding a firm publication deadline was reasonable when you were being warned of evidence that would materially contradict many claims within your post. When the reputation of someone within their own community is at stake, you want to dot every i and cross every t, and spending hundreds of hours over the course of several months collecting accusations and a single-digit number in a week collecting rebuttals, up to and including making significant corrections on the final day and rejecting other significant corrections as being too late to add, does not meet that standard.
Trying someone in the court of public opinion can be as serious a matter as trying them in the court of law, and while you say the goal of the post was not to judge, that is both hard to square with lines like “I expect that if Nonlinear does more hiring in the EA ecosystem it is more-likely-than-not to chew up and spit out other bright-eyed young EAs who want to do good in the world” and hard to square with the duty you take on by electing to publish an exposé about someone. You have to play the role of judge in an article like that, and you are not fulfilling your duty by circulating allegations without putting in sufficient due diligence.
In those hundreds of hours, so far as I can tell, you spent a grand total of three gathering one side of the story—the side of a group Ben’s post was explicitly adversarial towards and explicitly aimed at destroying the reputation of.
Look, man, what do you want us to do. We talked to dozens of other people working at the organization, extensively cross-checked the stories they told us, and also did extensive independent research to figure out what happened. Also, our sources were terrified of retribution and did not want to be exposed to the leadership.
We talked to them multiple times, for many hours, though yeah, we weren’t able to show them everything we were planning to publish. What is the threshold of hours of engagement where it’s OK to publish such an article according to your perspective?
This is all totally standard stuff when you do investigative reporting. Of course you don’t always give the organization you are reporting on full access to your article before you publish it. I wish people could reliably do that, but it’s genuinely hard with sources who are worried about retribution.
I encourage you to talk to any investigative reporter with experience in the field and ask them whether your demands here are at all realistic for anyone working in the space.
and spending hundreds of hours over the course of several months collecting accusations and a single-digit number in a week collecting rebuttals, up to and including making significant corrections on the final day
This is also really not an accurate summary. Most of the months were spent crossing the i’s and dotting the t’s. They were spent trying to find contradictory evidence, and they were also spent following up on concrete things that the Nonlinear leadership told us about.
I told you what I wanted you to do. You say it’s unrealistic, I say that I wouldn’t think to hold you to a standard I don’t hold myself to.
I’m a blogger and a podcast producer, not an investigative journalist, and I’m paid to focus on internet nonsense, not truly critical, world-saving stuff. But I do my fair share of investigative work. This article is one of my recent high-effort investigations, over someone getting punched on a beach after a long series of ugly subculture drama—a much lower-stakes sequence than the one you covered, but one no less personal or painful for the participants.
The “antagonists” were not particularly communicative, but I reached out to them multiple times, including right before publication, checking if I could ask questions and asking them to review my claims about them for accuracy. I went to the person closest to them who was informed on the situation and got as much information as I could from them. I spent hours talking with my primary sources, the victim and his boyfriend, and collecting as much hard evidence as possible. I spent a long time weighing which points were material and which would just serve to stir up and uncover old drama. Parties claimed I was making major material errors at several points during the process, and I dug into their claims as thoroughly as I could and asked for all available evidence to verify. Often, the disputes they claimed were material hinged on dissatisfaction with framing.
All sources were, mutually, worried about retribution and vitriol from the other parties involved. All sources were part of the same niche subculture spaces, all had interacted many times over the past half-decade, mostly unhappily, and all had complicated, ugly backstories.
I was not paid for this, except inasmuch as I’m paid a part-time monthly salary for podcast production work. I did it in my spare time while balancing a full law school schedule. I approached it with care, with seriousness, and with full understanding of the reputational effects I expected it to have and the evidence I had backing and justifying those effects. What I want you to do is exactly what I would do if I were assigned this task, given comparable timing and hour constraints. No more, no less.
There’s no threshold of hours of engagement. The test I am describing is this: are you receiving, or do you seem likely to receive, new material facts that contradict elements of your narrative? You were, up until two hours before publication, with a promise that there was more on the way. There is nothing unreasonable about saying publication should be delayed in that circumstance.
Anyway, look, I’m not an investigative reporter with experience in the field, much as I LARP as one online. That said, I’m on good personal terms with several and am happy to put my money where my mouth is and check with them. Let me know if the following is an appropriate summary or whether you’d make changes:
Say you were advising someone on a story they’d been working on for six months aimed at presenting an exposé of a group their sources were confident was doing harm. They’d contacted dozens of people, cross-checked stories, and did extensive independent research over the course of hundreds of hours, paying their key sources for the trouble. Their sources, who will be anonymous but realistically identifiable in the article, express serious concerns about retribution and request a precise, known-in-advance publication date. Towards the end, they contact the group in question, with their primary conversation lasting a few hours and coming less than a week before their scheduled publication date. The group claims that several points in their article are materially wrong and libelous and asks for another week to compile evidence to rebut those claims, growing increasingly frantic as the publication date approaches and escalating to a threat of a libel suit. On the last day before publication, they show a draft to another person close to the story, who makes at least one clear correction of material fact, then, with a couple of hours to go before the scheduled publication, provides evidence contradicting another of the article’s supporting claims.
Would you advise them to publish the article in its current form, make a last-minute edit to include the final piece of material evidence, wait another week to review the claimed rebuttal, or take some other course of action?
I like this summary, actually! Some small edits I would make:
On the last day before publication, they show a draft to another person close to the story, who makes at least one clear correction of material fact, then, with a couple of hours to go before the scheduled publication, provides evidence contradicting another of the article’s supporting claims.
This is currently inaccurate. Spencer did not point out a material inaccuracy. Here is the original quote that we showed Spencer:
This is all totally accurate. Spencer pointed out that a good chunk of the relevant review period was not the one in which Emerson was CEO, and so not all of these reviews must be of Emerson as CEO (which we did not claim, but I can see how someone might read it that way). So we edited it to be the following:
I think the previous reviews were relevant to the case at hand, and the evidence was accurate as presented (even if they were about the company that Emerson left behind, and not while he was there, though some of the reviews were about Emerson). We did nevertheless replace it with just reviews from the relevant time period, to make things more robust, but I do not consider the previous presentation inaccurate, and the pointer to the glassdoor page to still be relevant evidence (especially given the one review we quoted here, which mentions Spartz directly).
As such, it is not accurate to say that material inaccuracies were pointed out to us in the post.
Towards the end, they contact the group in question, with their primary conversation lasting a few hours and coming less than a week before their scheduled publication date.
This is also inaccurate in that we first talked to Nonlinear about the accusations many months before the publication date, when we weren’t sure yet whether we would publish anything on this topic. What is accurate is that we only informed Nonlinear that we are going to publish a post and informed them about the material facts.
Here is how I would currently phrase things, leaving most of your language intact:
Say you were advising someone on a story they’d been working on for six months aimed at presenting an exposé of a group their sources were confident was doing harm. They’d contacted dozens of people, cross-checked stories, and did extensive independent research over the course of hundreds of hours.
Their sources, who will be anonymous but realistically identifiable in the article, express serious concerns about retribution and request a known-in-advance publication date.
They have talked to the group they are investigating multiple times to gather evidence, but have not informed them that they are planning to release an exposé with the evidence they gathered. 7 days before their scheduled publication date they contact the group and inform them about their intent to publish and the key claims they are planning to include in their exposé.
The group claims that several points in their article are materially wrong and libelous and asks for another week to compile evidence to rebut those claims, growing increasingly frantic as the publication date approaches and escalating to a threat of a libel suit.
On the last day before publication, they show a draft to another person close to the story who points out a detail that does not directly contradict anything in the post, but seems indirectly implied to be false, which they correct in the final publication. Then with two hours to go before the scheduled publication, the same contact provides evidence against one of the statements made in the post, though also does not definitely disprove it.
Would you advise them to publish the article in its current form, or delay publication, despite the credible requests about the sources about retribution and the promise of the scheduled publication date?
If you would be up for sending something like this to someone who works in investigative journalism, I would actually appreciate it. Some things that seem relevant to clarify:
I would really like you to avoid framing the question too much. I think this is very easy among friends.
I think it would be good to differentiate “what he considers prudent for your own libel risk” from “what he considers his ethical responsibility”. I think it’s somewhat plausible (though not that likely) that he would be like “well, you really want to avoid a libel suit so you have to halt publication”, but I am more interested what he would consider ethical, ignoring self-interest (I think whether it was prudent to go ahead given libel risk is an interesting question, but not I hope to answer with this inquiry).
Thank you for offering to do this. I do find myself pretty interested in the answer. I don’t think an answer one way or another would totally flip me here, but I would definitely update somewhat.
It’s tempting to nitpick the edits a bit, but I think this is probably close enough to get good answers while being approximately satisfactory to both of us. I’ll let you know how it goes.
The above image contains the full text of my message, absent the rest of the copy-pasted hypo. I’ll note in the interests of broad fairness that other involved parties suggested edits, notably that the last-minute evidence was evidence indicating the key witness had lied and that “7 days” is longer than they had to respond to the material claims. I used none of their suggestions. I think the hypo could be a reasonable question across a somewhat broad range of specific factual emphases and think the framing as-is is sufficient to get good answers; in my messages, I did not alter the hypo from the words you chose.
I reached out to three journalists with long investigative track records and have two responses so far. It goes without saying that these are people I have close working relationships, regular communication, or other personal connections with, but I believe the framing and lack of context provided mean they are well-positioned to consider the question in the abstract and on the merits independent of any connections.
The first response (update: from Katie Herzog):
I would delay publication. I’m not sure about the specifics of libel law but putting myself in a publisher’s shoes, they do tend to not want to get sued and your first commitment, beyond getting the scoop or even stopping the hypothetical group from doing harm, should be towards accuracy.
The second (update: from Jesse Singal):
I think it depends a lot on the group’s ability to provide evidence the investigators’ claims are wrong. In a situation like that I would really press them on the specifics. They should be able to provide evidence fairly quickly. You don’t want a libel suit but you also don’t want to let them indefinitely delay the publication of an article that will be damaging to them. It is a tricky situation! I am not sure an investigative reporter would be able to help much more simply because what you’re providing is a pretty vague account, though I totally understand the reasons why that’s necessary.
UPDATE:
The third, from Helen Lewis:
This feels like a good example of why you shouldn’t over-promise to your sources—you want a cordial relationship with them but you need boundaries too. I can definitely see a situation where you would agree to give a source a heads up once you’d decided to publish — if it was a story where they’d recounted a violent incident or sexual assault, or if they needed notice to stay somewhere else or watch out for hacking attempts. But I would be very wary of agreeing in advance when I would publish an investigation—it isn’t done until it’s done.
In the end the story is going out under your name, and you will face the legal and ethical consequences, so you can’t publish until you’re satisfied. If the sources are desperate to make the information public, they can make a statement on social media. Working with a journalist involves a trade-off: in exchange for total control, you get greater credibility, plausible deniability and institutional legal protection. If I wasn’t happy with a story against a ticking clock, I wouldn’t be pressured into publication. That’s a huge risk of libelling the subjects of the piece and trashing your professional reputation.
On the request for more time for right to reply, that’s a judgement call—is this a fair period for the allegations involved, or time wasting? It’s not unknown for journalists to put in a right to reply on serious allegations, and the subject ask for more time, and then try to get ahead of the story by breaking it themselves (by denying it).
Hmm, the first one seems to be responding at the level of “here is how you don’t get sued”. Would be interested in a follow-up question asking what to do if you are not concerned about getting sued.
The second one makes sense. Would be happy to draft something with more detail as a response, so we can get something out of it.
I would change the text. He gave us less than 24 hours.
He sent us the draft in the middle of the night, filled with many accusations we hadn’t even heard of, on a day he knew we were traveling and wouldn’t be able to respond properly. He said he’d publish it that very day (aka <24 hours)
He ended up publishing it the next day at a time where we normally would have been asleep, except that we’d asked a friend to call us and wake us up if Ben was posting. We ended up having to respond to that post on a fraction of the sleep we usually get.
You could try to save he gave us 60 hours if you count from the time he spoke to us to the time he published. However, when we spoke to him, we thought he would wait to see our evidence. He also didn’t tell us many of the accusations he was going to publish, so I think this is an unfair characterization of the time they gave us
He did not promise to look at the evidence before publishing, so he was consistent in that regard, but we thought he would wait since he explicitly said in a follow up email: “FYI I did update from things you shared that Alice’s reports are less reliable than I had thought, and I do expect you’ll be able to show a bunch of the things you said.”
He did not wait to see the evidence. The evidence he’d already seen had, in his own words, made him realize that Alice was less reliable than he thought, and he knew we were sending him things like interview transcripts and screenshots providing concrete evidence that they’d told him falsehoods and misleading claims.
And he published anyways.
I think it would be good to share that with your journalist friends.
Here’s the relevant section explaining the whole timeline.
I’ve been trying to stay out of this, but I’m honestly shocked at this claim you’re making.
You say:
We talked to dozens of other people working at the organization, extensively cross-checked the stories they told us
But this is, just, wildly false? You did not speak to dozens of other people working at Nonlinear.
And Ben himself contradicts you. In Ben’s post, he says:
My current understanding is that they’ve had around ~4 remote interns, 1 remote employee, and 2 in-person employees (Alice and Chloe).
Ben thinks we’ve only had 7 total team members, but we’ve actually had 21 - extremely far off.
If you “extensively cross-checked the stories,” how did Ben get such a basic number so wrong? And why are you under the impression that you had talked to dozens of employees if Ben did not?
The fact that you spent 1000 hours on this and got such key details this wrong is surprising to me.
Ok, but why is this a big deal? Aside from showing egregiously bad fact checking, a large portion of Ben’s post was trying to make the case that there is a pattern of Nonlinear “chewing up and spitting out other bright-eyed young EAs who want to do good in the world.” It would significantly weaken your case if it were 2 out of 21 team members [1]were unhappy instead of 2 out of 7.
Not only that, but to my knowledge, Ben did not talk to a single employee or intern since Alice and Chloe to see if these patterns were, in fact, patterns.
Sorry, saying “worked at” is definitely not the right term, sorry about that.
We talked to dozens of people who have either worked at Nonlinear, otherwise worked with people currently at Nonlinear, or have substantially engaged with Nonlinear in a professional capacity and so seem like they are in a good position to judge what happened. “Worked at” is definitely the wrong word. I should have said something like “have worked with people at Nonlinear”.
I don’t particularly want to litigate the employee thing in this random thread. My best guess is Ben was talking about the number of employees during the specific stretch of months that the article was covering.
It is also inaccurate that only 2 employees we talked to had bad experiences. As Ben mentions multiple times in the post, many additional people we talked to had bad experiences (though generally of somewhat lesser magnitude).
This squares very starkly in contrast with Nonlinear’s perception of things. It seems to me that all the work in your comment is being done by the “we did not believe them on” bit, which is very subjective and frankly would be ridiculous in something like fair trial—it would be like saying “the defense is not allowed to bring witnesses or make a case, because despite them claiming that they’ll make a strong case, we (the prosection) just don’t believe them”. You can argue about whether Nonlinear’s eventual response was satisfactory (though their evidence seems compelling to me), but I’m not seeing your case on this point in particular.
Hmm, I don’t think I am understanding this comment, so might be best to just clarify.
Ben’s goal with the post was really not to be judge, I hope he made that abundantly clear. The goal was to publish some evidence that had been extensively circulating around privately in the EA Community for a while, so that more people could take it into account, and also allow Nonlinear to publish a response or try to refute that evidence.
For that purpose, the question is whether Ben published anything that he knew was wrong. He did not do so, to the best of my knowledge. Nonlinear objected to a bunch of stuff, and we tried our best to summarize their objections in the post (in the section that is a summary of Nonlinear’s position). Lightcone did not (and continues to not) have the capacity to fully validate every claim given to us, though like, we did spend close to a thousand hours in terms of talking to sources and trying to validate and fact-check the things that were given to us. This included talking to Nonlinear and engaging with their evidence, though we were quite limited in what things our sources allowed us to share with them before publication.
Given this, I am not really sure what point you are making in this comment. I think it would have been bad for us to selectively fail to publish evidence that we did have, but we did not do so (though Nonlinear is accusing Ben of doing that, which is false, as far as I can tell). I agree we could have of course spent more time fact-checking things, but again we were limited in what we could share with Nonlinear before publication, and also had already spent hundreds of hours doing that kind of work, conducting interviews with over a dozen different people who had experiences with Nonlinear, cross-checking various details and facts to verify to the best of our ability that they added up.
Hm, a lot I disagree with here, but a crux is that I think you’re not really replying to TracingWoodgrain’s original point, which was that Ben knew there might be significant evidence contradicting much of his post but decided not to wait for it and published anyways (which TW considers to be a bad norm). Instead you seem to be changing frame to “did Ben publish anything which he knew for sure wasn’t true”, which is quite different, particularly in this case where evidence is deliberately not being looked at.
Ah, sorry, I did understand your question to be about the latter. That’s just a relatively straightforward misunderstanding. Might write more on the former.
Spencer sent us a screenshot about the vegan food stuff 2 hours before publication, which Ben didn’t get around to editing in before the post went live, but that’s all the evidence that I know about that you could somehow argue we had but didn’t include. It is not accurate that Nonlinear sent credible signals of having counterevidence before the post went live
Uh, actually I do think that being sent screenshots showing that claims made in the post are false 2 hours before publication is a credible signal that Nonlinear has counterevidence.
I can’t believe I’m saying this, but I’m currently leaning towards the position that Lightcone deserves to be sued for defamation. Maybe not for “maximum damages permitted by law” (since those are truly excessive), but you probably owe them significant material compensation.
Uh, actually I do think that being sent screenshots showing that claims made in the post are false 2 hours before publication is a credible signal that Nonlinear has counterevidence.
Sorry, can you please explain what you would have liked us to do at this point? It’s 2 hours to publication, which is a major undergoing basically launching something that has been worked on for hundreds of hours.
The screenshots relate to one claim in a post with many dozens of claims, and do not directly falsify what is said in the post, but seem to relate to them in a somewhat complicated manner (see this discussion on the post). We are getting dozens of calls by Nonlinear who are, from our perspective, using a bunch of really quite aggressive tactics to prevent publication of this post at the same time.
Please specify concretely what you would have liked us to do instead? Completely halt publication of the post, against the direct promises we made to our sources, who have shown us credible evidence that they are worried about retaliation? I think the right thing to do is to leave a comment with the evidence, which we were indeed going to do if Kat hadn’t already done that within an hour of publication of the post.
Please be concrete what you would have liked us to do instead? I don’t think the screenshots were some kind of major smoking gun or whatever, they were a piece of evidence that was definitely related to one of the claims, but definitely not the kind of thing that would cause me to immediately update and throw out or delay the whole post.
Concretely, here’s what you should have done instead:
Pause. Wait. Delay. Completely halt publication of the post, against the direct promises you made to your sources, who have shown you credible evidence that they are worried about retaliation. Make your excuses, make your apologies, recognize that you are on the verge of publishing a partial narrative aimed at dealing serious reputational damage to someone in your sphere, and take a moment to absolutely ensure all the i’s are dotted and all the t’s are crossed.
If you receive concrete evidence that you are about to publish a falsehood aimed at inflicting material harm on someone, your duty is simply not to do that. If you are scrambling on the last day to update clear material falsehoods in your post, as Spencer reports with the company reviews, something has failed in your fact-checking process, and you should grimace, take a deep breath, and pause publication until you have things figured out.
Of course Nonlinear was using aggressive tactics. There’s no truly returning from reputational damage inflicted by a post of that sort. Even if a response sways some people, it will at best always result in a sharp community divide, with people taking sides and landing with one adversarial party (you) or the other (them). I was unfamiliar with Nonlinear and Lightcone until these events, and don’t have any sort of personal stake in the fight other than a desire to see careful standards applied to investigative work. But it seems abundantly clear to me that proper caution was not taken in the publication of the original post, and the root of that seems to be an impression that it’s alright to publish material falsehoods if you update later, and that it’s more important to keep to a publication schedule than to ensure you’re presenting a complete picture. I emphatically disagree on both counts.
I think the approximate result of this tactic is that we would have never been able to publish such a post (because Nonlinear would have been capable of producing an infinite trickle of relevant-seeming evidence that requires extensive investigation to thoroughly debunk), and that our sources would have been heavily retaliated against and would have regretted talking to us in the first place.
You are free to argue that such posts should approximately never be written, and I am interested in that argument. But I think our willingness to pay to make any kind of investigation in this space happen, and the diligence with which we went about this, including on the dimension of gathering contradicting evidence, is a vast and far outlier for stuff like this.
If you want anyone to do this kind of thing, you need to be OK with a lower standard for information propagation. And maybe you just don’t want posts like this, that’s OK. I do think this kind of post is really important and crucial and I desperately wish there were more of them.
Nonlinear would have been capable of producing an infinite trickle of relevant-seeming evidence
It reads to me that NL was asking for a specific amount of time to provide you with evidence, and you could have prevented this turning into unbounded delay by saying something like “you’ve asked for a week to gather evidence, and we’ll consider evidence you provide by [168hr in the future]”.
our sources would have been heavily retaliated against
It reads to me that NL was asking for a specific amount of time to provide you with evidence, and you could have prevented this turning into unbounded delay by saying something like “you’ve asked for a week to gather evidence, and we’ll consider evidence you provide by [168hr in the future]”.
Yeah, I am sorry, I think I was exaggerating the alternative here.
I do think we kind of did this with our original email and set of calls to them, and then they asked us for an additional week, and we said no. I think it’s pretty plausible we should have waited an additional week, and I also felt really conflicted and sad about this at the time. I do think it’s not reasonable to treat declining a request for an additional week as some kind of major and extreme faux pas, especially given the threat of libel and other types of what still seem to me to be undue forms of pressure that were applied.
So in that week Nonlinear sends a giant pile of evidence, which I expect I would not have found very compelling (though I am sure it would have probably been capable of falsifying some relatively minor points).
I do think we kind of did this with our original email and set of calls to them, and then they asked us for an additional week, and we said no.
How much of the claims made in the post did you share with them in the original email (or soon after, if you want to start the “clock” then) vs later?
(It sounds to me like there were many claims they believed were false that weren’t shared with them until quite close to publication, which matters quite a bit for interpreting the request for a week to provide evidence)
((Your phrasing of “additional week” here and above is weird because it implies they had already had a week during which they knew the post’s claims and could be providing conflicting evidence, which I think you’re not claiming.))
We shared a 2000 word high-level summary of all important claims made in the original post, as far as I can tell. I sadly can’t share it in-full here because a bunch of the information would be deanonymizing, but I would be happy to set up a call and show you the email, or answer any specific questions you have about what evidence was shared.
Here is an overview including a bunch of quotes of what we sent Nonlinear then (and if you think this is important I can also fully remove all deanonymizing information from the email, which would take me on the order of an hour or two, and share it with you):
Basic information about when Nonlinear was founded and who worked there during the relevant period
Basic details about when Alice and Chloe worked there
That Chloe’s salary was verbally agreed to be $75k/year, with $1k/month in stipend and the rest in food/board/travel, which ultimately (according to Chloe and Alice) did not actually add up to $75k/year
That Alice joined as the sole person in the incubation program, and that she received no salary for the first months of traveling with Nonlinear
That Alice often had less than $1000 in her bank account and would often rack up substantial expenses in reimbursements, and that at the end of her employment Nonlinear owed her multiple thousands of dollars
There were no written specifications of how healthcare and covered medical bills would work
The story with Emerson and Adorian Deck and how it seemed very adversarial
Other rumors about Emerson doing questionably legal things in order to intimidate people he was in conflict with
As I understand it they were considered to have the lowest monetary value of time in the house, and consequently were given a lot of the menial tasks around the house.
“They also reported being strongly discouraged from spending time away from the Nonlinear house (e.g. living in a separate AirBnb) and with people that you folks didn’t consider valuable/worthwhile. They both reported very strong senses of social and financial dependence while working with you.”
“During her time at Nonlinear, [Alice] quit being a vegan. This was during a time when she caught Covid, and was unable to get enough vegan food to eat for ~2 days.
“You asked [Chloe] to do a lot of driving regularly for her job, but she didn’t know how to drive. You gave her driving lessons (I believe that was from Drew), and then encouraged her to drive without a license for 1-2 months. She eventually was freaked out about this and stopped and regrets doing it, and thinks that she would not have done so if she wasn’t in such an otherwise isolating and dependent environment.”
“There was a long period of difficult relations with [Alice] and Kat, and some conflict regarding monogamy/polyamory, that disrupted working together quite a bit.”
“Relatedly, as [Alice] was returning from Mexico, Kat made a request for her to bring back several drugs over the border, some recreational, some for productivity.”
“Both [Alice] and [Chloe] reported feeling really hurt by their time at Nonlinear, and taking some months to recover before they could go back to work.”
“[Alice] reports a lot of effusive positive emotions from you, including various familial and sometimes romantic feelings of love, yet also felt threatened in various ways about her career and that she was in some pretty difficult financial circumstances as a result of working for you. These things seem pretty incongruent to me and potentially quite manipulative.”
“I’m not sure how to quickly summarize this. It seems to me they had a lot of strongly negative experiences or made decisions they regret that you (their employers and managers and the majority of their social environment) have some substantial responsibility for, in setting up this environment.”
“A number of people I spoke to were concerned about retaliation from you if they shared their experiences with me, and initially only did so under condition of strict confidentiality, which I was willing to offer because I had a bunch of warning flags raised from various people I trust who reached out to me and also from the financial situation Kat reported when we spoke at Lightcone.”
“[Alice] showed me some texts from Kat that offered support with “basic survival stuff” like housing, if Alice would “commit to not saying bad things about us to them”. This sort of behavior makes it very hard for me to trust impressions I get about your team — if people with negative information are being given strong reasons to keep it quiet, then I can’t find out about negative information about you, and I have a lot of uncertainty about how much I don’t know. A sense that people were scared to share negative info, and concerns about covering-up negative info, are the main reason I’m trying to find out what happened with [Alice] and [Chloe] and others as much as I am.
“[Alice] also showed me texts from Kat saying that [Alice] was saying bad things about you behind your back, and said that if you did the same to her her career in EA “would be over in a few Dm’s”. I’m interested to know what the content of those hypothetical DM’s would be. To be clear this definitely reads to me as a veiled threat.
I’m confused by the heavy disagree-voting on this comment. Are people saying with their votes that this does not actually cover all the important claims in the post? If someone who disagree-voted (or otherwise disagrees) wanted to comment I’d find that illuminating!
One of habryka’s other comments in this thread suggests that someone (or some small group of people, or some single person with sockpuppets) is downvoting literally everything habryka posts in this thread.
It would be nice if there were some way for someone with no conflict of interest to investigate that...
Where are you getting the idea that Ben gave us a week? The draft was sent to us on the same day Ben said he was going to publish it. On a day Ben knew we were traveling and wouldn’t be able to respond properly (sketchy/no internet, chaos of traveling, etc).
We spoke to Ben 60 hours before he published, and he only told us a subset of all the accusations. A quick re-reading of the post and I found 14 allegations that were new that Ben hadn’t discussed on the call. And I only got a short way through re-reading the post (maybe a 20%?) because I find reading it extremely painful.
I posted an approximately full list of claims you were informed of 4 days before publishing.
Would you be able to highlight any important claims that were not included in that list of claims? It is totally possible there are some, but having cross-check the two, I can’t find any major ones.
I found 14 allegations that were new that Ben hadn’t discussed on the call. [in the first ~20%]
Were these fourteen included in the email Ben sent ~5d out (that Habryka summarizes here) and just not discussed on the call, or are you saying that the fourteen were first introduced to you with the final draft?
Strongly positive/negative relative to what? Relative to being more accurate initially, sure. Relative to being wrong but just not acknowledging it, no.
Specifically, I think people by default read this post as an example of good epistemic practice in a community built around good epistemic practice. In this case, though, I think the prior bad epistemic practice (not waiting for full information before publishing a highly consequential piece aimed at inflicting reputational damage on someone) is significant enough and bad enough that emphasizing a plan to update after the fact should be viewed primarily through the lens of damage control.
The standard with this sort of investigative piece should be to gather information of this nature prior to publication to whatever extent possible, where updating on new information after-the-fact is praiseworthy only if that information was not realistically knowable prior to publication.
You’re correct that it’s better than being wrong but not acknowledging it, but I think that’s a well-established standard in this sphere and there’s a stronger need to update based on the importance of gathering relevant information prior to publication.
emphasizing a plan to update after the fact should be viewed primarily through the lens of damage control.
Is anyone acting like that is not a damage control measure? I upvoted specifically because “do damage control” is better than “don’t”. Usually when I see a hit piece, and later there are a bunch of inaccuracies that come to light, I don’t in fact see that damage control done afterwards.
Also I think this kind of within-tribe conflict gets lots of attention within the EA and LW social sphere. I expect that if Ben publishes corrections a bunch of people will read them.
My read is that many people still consider the publication of the original post to be prudent and responsible given the circumstances, while any updates based on information that comes to light here will be prudent and responsible given the new information. Instead, I think people should view the original post as imprudent and irresponsible to the extent that it did not give one side of an adversarial situation an adequate hearing-out (and it really seems like it didn’t: a three-hour phone call where you misleadingly summarize their response as “Good summary!”, then refuse to wait until they can provide a more substantive response, is extraordinarily bad practice given the hundreds of hours he mentions putting into the rest of the investigation), with any subsequent updates being judged as returning towards responsibility after the fact rather than continuing a pattern of prudence.
I think it’s worth explicitly stating that while I think most in this community are reading “I plan to compare the things that this post says directly against specific claims in mine, and acknowledge anything where I was factually inaccurate” as an example of a positive norm, it reads to me as a strongly negative one.
Specifically: Based on this comment, it sounds like you knew or should have known much of this evidence was available upon the decision to publish your original post, and you went ahead regardless without waiting to review claims that falsify parts of your original post. In other words, you mixed what appear to be materially false claims with true ones in a post aimed explicitly at destroying the reputation of an organization. The correct time to update factual inaccuracies in something so high-stakes is not after-the-fact, it is prior to publication, and the presence of correctable falsehoods in the original, whether or not you subsequently correct them, is a major issue.
To be clear, as I think I have stated in like dozens of places, at the time the post was written, all evidence that Ben was given made it into the post. There are no glaring omissions, the post accurately captured Ben’s epistemic state at the time.
The only thing happening was that Nonlinear was claiming many of the concrete claims were false, which we did not believe them on (and I still don’t believe them on). Spencer sent us a screenshot about the vegan food stuff 2 hours before publication, which Ben didn’t get around to editing in before the post went live, but that’s all the evidence that I know about that you could somehow argue we had but didn’t include. It is not accurate that Nonlinear sent credible signals of having counterevidence before the post went live (and if someone wants to, I would probably be up for arranging a call with a neutral third party to walk them through the totality of our Nonlinear communications and see whether they find anything that really seems like it had to be included).
(The one other thing that didn’t end up in the post was Alice’s side-business, which we did know about but didn’t consider very important, and Nonlinear didn’t bring it up in the pre-publication conversations we had. Ben is sad about not including it, but it also didn’t seem that material to the case given that it probably never made much money, though me and Ben still think it would have been good to include. However, again, Nonlinear did not poke us about this or share evidence that it was important or was an important omission before publishing)
I responded to your EA forum comments much earlier in the day, but my responses are still stuck in the mod filter, so I might repeat myself a bit here. I’m not part of EA, but I’ve been around the rationalist sphere for a long while and I have more experience than most here with investigative journalism–adjacent work.
You regularly mention hundreds of hours of work, a deadline you had to meet, the inconvenience you would have faced upon delaying, and the chance that you might have chosen not to publish at all had you delayed.
That’s all very well and good, but honestly doesn’t sway the needle for me at all in terms of your duty in the situation. Self-imposed deadlines are not an excuse to avoid hearing out one party to a conflict you’re publishing.
In those hundreds of hours, so far as I can tell, you spent a grand total of three gathering one side of the story—the side of a group Ben’s post was explicitly adversarial towards and explicitly aimed at destroying the reputation of. When they told him they were compiling a detailed list of evidence to respond to some of his claims, you guys flatly refused to investigate further or delay your self-imposed deadline. I cover a story like this almost every week for the podcast I work for, and I put more time into gathering opposing stories than that.
I believe you when you say that (almost) all evidence Ben was given made it into the post. I emphatically disagree that holding a firm publication deadline was reasonable when you were being warned of evidence that would materially contradict many claims within your post. When the reputation of someone within their own community is at stake, you want to dot every i and cross every t, and spending hundreds of hours over the course of several months collecting accusations and a single-digit number in a week collecting rebuttals, up to and including making significant corrections on the final day and rejecting other significant corrections as being too late to add, does not meet that standard.
Trying someone in the court of public opinion can be as serious a matter as trying them in the court of law, and while you say the goal of the post was not to judge, that is both hard to square with lines like “I expect that if Nonlinear does more hiring in the EA ecosystem it is more-likely-than-not to chew up and spit out other bright-eyed young EAs who want to do good in the world” and hard to square with the duty you take on by electing to publish an exposé about someone. You have to play the role of judge in an article like that, and you are not fulfilling your duty by circulating allegations without putting in sufficient due diligence.
Look, man, what do you want us to do. We talked to dozens of other people working at the organization, extensively cross-checked the stories they told us, and also did extensive independent research to figure out what happened. Also, our sources were terrified of retribution and did not want to be exposed to the leadership.
We talked to them multiple times, for many hours, though yeah, we weren’t able to show them everything we were planning to publish. What is the threshold of hours of engagement where it’s OK to publish such an article according to your perspective?
This is all totally standard stuff when you do investigative reporting. Of course you don’t always give the organization you are reporting on full access to your article before you publish it. I wish people could reliably do that, but it’s genuinely hard with sources who are worried about retribution.
I encourage you to talk to any investigative reporter with experience in the field and ask them whether your demands here are at all realistic for anyone working in the space.
This is also really not an accurate summary. Most of the months were spent crossing the i’s and dotting the t’s. They were spent trying to find contradictory evidence, and they were also spent following up on concrete things that the Nonlinear leadership told us about.
I told you what I wanted you to do. You say it’s unrealistic, I say that I wouldn’t think to hold you to a standard I don’t hold myself to.
I’m a blogger and a podcast producer, not an investigative journalist, and I’m paid to focus on internet nonsense, not truly critical, world-saving stuff. But I do my fair share of investigative work. This article is one of my recent high-effort investigations, over someone getting punched on a beach after a long series of ugly subculture drama—a much lower-stakes sequence than the one you covered, but one no less personal or painful for the participants.
The “antagonists” were not particularly communicative, but I reached out to them multiple times, including right before publication, checking if I could ask questions and asking them to review my claims about them for accuracy. I went to the person closest to them who was informed on the situation and got as much information as I could from them. I spent hours talking with my primary sources, the victim and his boyfriend, and collecting as much hard evidence as possible. I spent a long time weighing which points were material and which would just serve to stir up and uncover old drama. Parties claimed I was making major material errors at several points during the process, and I dug into their claims as thoroughly as I could and asked for all available evidence to verify. Often, the disputes they claimed were material hinged on dissatisfaction with framing.
All sources were, mutually, worried about retribution and vitriol from the other parties involved. All sources were part of the same niche subculture spaces, all had interacted many times over the past half-decade, mostly unhappily, and all had complicated, ugly backstories.
I was not paid for this, except inasmuch as I’m paid a part-time monthly salary for podcast production work. I did it in my spare time while balancing a full law school schedule. I approached it with care, with seriousness, and with full understanding of the reputational effects I expected it to have and the evidence I had backing and justifying those effects. What I want you to do is exactly what I would do if I were assigned this task, given comparable timing and hour constraints. No more, no less.
There’s no threshold of hours of engagement. The test I am describing is this: are you receiving, or do you seem likely to receive, new material facts that contradict elements of your narrative? You were, up until two hours before publication, with a promise that there was more on the way. There is nothing unreasonable about saying publication should be delayed in that circumstance.
Anyway, look, I’m not an investigative reporter with experience in the field, much as I LARP as one online. That said, I’m on good personal terms with several and am happy to put my money where my mouth is and check with them. Let me know if the following is an appropriate summary or whether you’d make changes:
I like this summary, actually! Some small edits I would make:
This is currently inaccurate. Spencer did not point out a material inaccuracy. Here is the original quote that we showed Spencer:
This is all totally accurate. Spencer pointed out that a good chunk of the relevant review period was not the one in which Emerson was CEO, and so not all of these reviews must be of Emerson as CEO (which we did not claim, but I can see how someone might read it that way). So we edited it to be the following:
I think the previous reviews were relevant to the case at hand, and the evidence was accurate as presented (even if they were about the company that Emerson left behind, and not while he was there, though some of the reviews were about Emerson). We did nevertheless replace it with just reviews from the relevant time period, to make things more robust, but I do not consider the previous presentation inaccurate, and the pointer to the glassdoor page to still be relevant evidence (especially given the one review we quoted here, which mentions Spartz directly).
As such, it is not accurate to say that material inaccuracies were pointed out to us in the post.
This is also inaccurate in that we first talked to Nonlinear about the accusations many months before the publication date, when we weren’t sure yet whether we would publish anything on this topic. What is accurate is that we only informed Nonlinear that we are going to publish a post and informed them about the material facts.
Here is how I would currently phrase things, leaving most of your language intact:
If you would be up for sending something like this to someone who works in investigative journalism, I would actually appreciate it. Some things that seem relevant to clarify:
I would really like you to avoid framing the question too much. I think this is very easy among friends.
I think it would be good to differentiate “what he considers prudent for your own libel risk” from “what he considers his ethical responsibility”. I think it’s somewhat plausible (though not that likely) that he would be like “well, you really want to avoid a libel suit so you have to halt publication”, but I am more interested what he would consider ethical, ignoring self-interest (I think whether it was prudent to go ahead given libel risk is an interesting question, but not I hope to answer with this inquiry).
Thank you for offering to do this. I do find myself pretty interested in the answer. I don’t think an answer one way or another would totally flip me here, but I would definitely update somewhat.
It’s tempting to nitpick the edits a bit, but I think this is probably close enough to get good answers while being approximately satisfactory to both of us. I’ll let you know how it goes.
Thank you, I appreciate it!
The above image contains the full text of my message, absent the rest of the copy-pasted hypo. I’ll note in the interests of broad fairness that other involved parties suggested edits, notably that the last-minute evidence was evidence indicating the key witness had lied and that “7 days” is longer than they had to respond to the material claims. I used none of their suggestions. I think the hypo could be a reasonable question across a somewhat broad range of specific factual emphases and think the framing as-is is sufficient to get good answers; in my messages, I did not alter the hypo from the words you chose.
I reached out to three journalists with long investigative track records and have two responses so far. It goes without saying that these are people I have close working relationships, regular communication, or other personal connections with, but I believe the framing and lack of context provided mean they are well-positioned to consider the question in the abstract and on the merits independent of any connections.
The first response (update: from Katie Herzog):
The second (update: from Jesse Singal):
UPDATE:
The third, from Helen Lewis:
Thank you!
Hmm, the first one seems to be responding at the level of “here is how you don’t get sued”. Would be interested in a follow-up question asking what to do if you are not concerned about getting sued.
The second one makes sense. Would be happy to draft something with more detail as a response, so we can get something out of it.
I would change the text. He gave us less than 24 hours.
He sent us the draft in the middle of the night, filled with many accusations we hadn’t even heard of, on a day he knew we were traveling and wouldn’t be able to respond properly. He said he’d publish it that very day (aka <24 hours)
He ended up publishing it the next day at a time where we normally would have been asleep, except that we’d asked a friend to call us and wake us up if Ben was posting. We ended up having to respond to that post on a fraction of the sleep we usually get.
You could try to save he gave us 60 hours if you count from the time he spoke to us to the time he published. However, when we spoke to him, we thought he would wait to see our evidence. He also didn’t tell us many of the accusations he was going to publish, so I think this is an unfair characterization of the time they gave us
He did not promise to look at the evidence before publishing, so he was consistent in that regard, but we thought he would wait since he explicitly said in a follow up email: “FYI I did update from things you shared that Alice’s reports are less reliable than I had thought, and I do expect you’ll be able to show a bunch of the things you said.”
He did not wait to see the evidence. The evidence he’d already seen had, in his own words, made him realize that Alice was less reliable than he thought, and he knew we were sending him things like interview transcripts and screenshots providing concrete evidence that they’d told him falsehoods and misleading claims.
And he published anyways.
I think it would be good to share that with your journalist friends.
Here’s the relevant section explaining the whole timeline.
I’ve been trying to stay out of this, but I’m honestly shocked at this claim you’re making.
You say:
But this is, just, wildly false? You did not speak to dozens of other people working at Nonlinear.
And Ben himself contradicts you. In Ben’s post, he says:
Ben thinks we’ve only had 7 total team members, but we’ve actually had 21 - extremely far off.
If you “extensively cross-checked the stories,” how did Ben get such a basic number so wrong? And why are you under the impression that you had talked to dozens of employees if Ben did not?
The fact that you spent 1000 hours on this and got such key details this wrong is surprising to me.
Ok, but why is this a big deal? Aside from showing egregiously bad fact checking, a large portion of Ben’s post was trying to make the case that there is a pattern of Nonlinear “chewing up and spitting out other bright-eyed young EAs who want to do good in the world.” It would significantly weaken your case if it were 2 out of 21 team members [1]were unhappy instead of 2 out of 7.
Not only that, but to my knowledge, Ben did not talk to a single employee or intern since Alice and Chloe to see if these patterns were, in fact, patterns.
This seems like poor truth-seeking to me.
edit: changed “employees” to “team members”
Sorry, saying “worked at” is definitely not the right term, sorry about that.
We talked to dozens of people who have either worked at Nonlinear, otherwise worked with people currently at Nonlinear, or have substantially engaged with Nonlinear in a professional capacity and so seem like they are in a good position to judge what happened. “Worked at” is definitely the wrong word. I should have said something like “have worked with people at Nonlinear”.
I don’t particularly want to litigate the employee thing in this random thread. My best guess is Ben was talking about the number of employees during the specific stretch of months that the article was covering.
It is also inaccurate that only 2 employees we talked to had bad experiences. As Ben mentions multiple times in the post, many additional people we talked to had bad experiences (though generally of somewhat lesser magnitude).
This squares very starkly in contrast with Nonlinear’s perception of things. It seems to me that all the work in your comment is being done by the “we did not believe them on” bit, which is very subjective and frankly would be ridiculous in something like fair trial—it would be like saying “the defense is not allowed to bring witnesses or make a case, because despite them claiming that they’ll make a strong case, we (the prosection) just don’t believe them”. You can argue about whether Nonlinear’s eventual response was satisfactory (though their evidence seems compelling to me), but I’m not seeing your case on this point in particular.
Hmm, I don’t think I am understanding this comment, so might be best to just clarify.
Ben’s goal with the post was really not to be judge, I hope he made that abundantly clear. The goal was to publish some evidence that had been extensively circulating around privately in the EA Community for a while, so that more people could take it into account, and also allow Nonlinear to publish a response or try to refute that evidence.
For that purpose, the question is whether Ben published anything that he knew was wrong. He did not do so, to the best of my knowledge. Nonlinear objected to a bunch of stuff, and we tried our best to summarize their objections in the post (in the section that is a summary of Nonlinear’s position). Lightcone did not (and continues to not) have the capacity to fully validate every claim given to us, though like, we did spend close to a thousand hours in terms of talking to sources and trying to validate and fact-check the things that were given to us. This included talking to Nonlinear and engaging with their evidence, though we were quite limited in what things our sources allowed us to share with them before publication.
Given this, I am not really sure what point you are making in this comment. I think it would have been bad for us to selectively fail to publish evidence that we did have, but we did not do so (though Nonlinear is accusing Ben of doing that, which is false, as far as I can tell). I agree we could have of course spent more time fact-checking things, but again we were limited in what we could share with Nonlinear before publication, and also had already spent hundreds of hours doing that kind of work, conducting interviews with over a dozen different people who had experiences with Nonlinear, cross-checking various details and facts to verify to the best of our ability that they added up.
Hm, a lot I disagree with here, but a crux is that I think you’re not really replying to TracingWoodgrain’s original point, which was that Ben knew there might be significant evidence contradicting much of his post but decided not to wait for it and published anyways (which TW considers to be a bad norm). Instead you seem to be changing frame to “did Ben publish anything which he knew for sure wasn’t true”, which is quite different, particularly in this case where evidence is deliberately not being looked at.
Ah, sorry, I did understand your question to be about the latter. That’s just a relatively straightforward misunderstanding. Might write more on the former.
Uh, actually I do think that being sent screenshots showing that claims made in the post are false 2 hours before publication is a credible signal that Nonlinear has counterevidence.
I can’t believe I’m saying this, but I’m currently leaning towards the position that Lightcone deserves to be sued for defamation. Maybe not for “maximum damages permitted by law” (since those are truly excessive), but you probably owe them significant material compensation.
Sorry, can you please explain what you would have liked us to do at this point? It’s 2 hours to publication, which is a major undergoing basically launching something that has been worked on for hundreds of hours.
The screenshots relate to one claim in a post with many dozens of claims, and do not directly falsify what is said in the post, but seem to relate to them in a somewhat complicated manner (see this discussion on the post). We are getting dozens of calls by Nonlinear who are, from our perspective, using a bunch of really quite aggressive tactics to prevent publication of this post at the same time.
Please specify concretely what you would have liked us to do instead? Completely halt publication of the post, against the direct promises we made to our sources, who have shown us credible evidence that they are worried about retaliation? I think the right thing to do is to leave a comment with the evidence, which we were indeed going to do if Kat hadn’t already done that within an hour of publication of the post.
Please be concrete what you would have liked us to do instead? I don’t think the screenshots were some kind of major smoking gun or whatever, they were a piece of evidence that was definitely related to one of the claims, but definitely not the kind of thing that would cause me to immediately update and throw out or delay the whole post.
Concretely, here’s what you should have done instead:
Pause. Wait. Delay. Completely halt publication of the post, against the direct promises you made to your sources, who have shown you credible evidence that they are worried about retaliation. Make your excuses, make your apologies, recognize that you are on the verge of publishing a partial narrative aimed at dealing serious reputational damage to someone in your sphere, and take a moment to absolutely ensure all the i’s are dotted and all the t’s are crossed.
If you receive concrete evidence that you are about to publish a falsehood aimed at inflicting material harm on someone, your duty is simply not to do that. If you are scrambling on the last day to update clear material falsehoods in your post, as Spencer reports with the company reviews, something has failed in your fact-checking process, and you should grimace, take a deep breath, and pause publication until you have things figured out.
Of course Nonlinear was using aggressive tactics. There’s no truly returning from reputational damage inflicted by a post of that sort. Even if a response sways some people, it will at best always result in a sharp community divide, with people taking sides and landing with one adversarial party (you) or the other (them). I was unfamiliar with Nonlinear and Lightcone until these events, and don’t have any sort of personal stake in the fight other than a desire to see careful standards applied to investigative work. But it seems abundantly clear to me that proper caution was not taken in the publication of the original post, and the root of that seems to be an impression that it’s alright to publish material falsehoods if you update later, and that it’s more important to keep to a publication schedule than to ensure you’re presenting a complete picture. I emphatically disagree on both counts.
I think the approximate result of this tactic is that we would have never been able to publish such a post (because Nonlinear would have been capable of producing an infinite trickle of relevant-seeming evidence that requires extensive investigation to thoroughly debunk), and that our sources would have been heavily retaliated against and would have regretted talking to us in the first place.
You are free to argue that such posts should approximately never be written, and I am interested in that argument. But I think our willingness to pay to make any kind of investigation in this space happen, and the diligence with which we went about this, including on the dimension of gathering contradicting evidence, is a vast and far outlier for stuff like this.
If you want anyone to do this kind of thing, you need to be OK with a lower standard for information propagation. And maybe you just don’t want posts like this, that’s OK. I do think this kind of post is really important and crucial and I desperately wish there were more of them.
It reads to me that NL was asking for a specific amount of time to provide you with evidence, and you could have prevented this turning into unbounded delay by saying something like “you’ve asked for a week to gather evidence, and we’ll consider evidence you provide by [168hr in the future]”.
Have you elaborated somewhere on this?
Yeah, I am sorry, I think I was exaggerating the alternative here.
I do think we kind of did this with our original email and set of calls to them, and then they asked us for an additional week, and we said no. I think it’s pretty plausible we should have waited an additional week, and I also felt really conflicted and sad about this at the time. I do think it’s not reasonable to treat declining a request for an additional week as some kind of major and extreme faux pas, especially given the threat of libel and other types of what still seem to me to be undue forms of pressure that were applied.
So in that week Nonlinear sends a giant pile of evidence, which I expect I would not have found very compelling (though I am sure it would have probably been capable of falsifying some relatively minor points).
How much of the claims made in the post did you share with them in the original email (or soon after, if you want to start the “clock” then) vs later?
(It sounds to me like there were many claims they believed were false that weren’t shared with them until quite close to publication, which matters quite a bit for interpreting the request for a week to provide evidence)
((Your phrasing of “additional week” here and above is weird because it implies they had already had a week during which they knew the post’s claims and could be providing conflicting evidence, which I think you’re not claiming.))
We shared a 2000 word high-level summary of all important claims made in the original post, as far as I can tell. I sadly can’t share it in-full here because a bunch of the information would be deanonymizing, but I would be happy to set up a call and show you the email, or answer any specific questions you have about what evidence was shared.
Here is an overview including a bunch of quotes of what we sent Nonlinear then (and if you think this is important I can also fully remove all deanonymizing information from the email, which would take me on the order of an hour or two, and share it with you):
Basic information about when Nonlinear was founded and who worked there during the relevant period
Basic details about when Alice and Chloe worked there
That Chloe’s salary was verbally agreed to be $75k/year, with $1k/month in stipend and the rest in food/board/travel, which ultimately (according to Chloe and Alice) did not actually add up to $75k/year
That Alice joined as the sole person in the incubation program, and that she received no salary for the first months of traveling with Nonlinear
That Alice often had less than $1000 in her bank account and would often rack up substantial expenses in reimbursements, and that at the end of her employment Nonlinear owed her multiple thousands of dollars
There were no written specifications of how healthcare and covered medical bills would work
The story with Emerson and Adorian Deck and how it seemed very adversarial
Other rumors about Emerson doing questionably legal things in order to intimidate people he was in conflict with
As I understand it they were considered to have the lowest monetary value of time in the house, and consequently were given a lot of the menial tasks around the house.
“They also reported being strongly discouraged from spending time away from the Nonlinear house (e.g. living in a separate AirBnb) and with people that you folks didn’t consider valuable/worthwhile. They both reported very strong senses of social and financial dependence while working with you.”
“During her time at Nonlinear, [Alice] quit being a vegan. This was during a time when she caught Covid, and was unable to get enough vegan food to eat for ~2 days.
“You asked [Chloe] to do a lot of driving regularly for her job, but she didn’t know how to drive. You gave her driving lessons (I believe that was from Drew), and then encouraged her to drive without a license for 1-2 months. She eventually was freaked out about this and stopped and regrets doing it, and thinks that she would not have done so if she wasn’t in such an otherwise isolating and dependent environment.”
“There was a long period of difficult relations with [Alice] and Kat, and some conflict regarding monogamy/polyamory, that disrupted working together quite a bit.”
“Relatedly, as [Alice] was returning from Mexico, Kat made a request for her to bring back several drugs over the border, some recreational, some for productivity.”
“Both [Alice] and [Chloe] reported feeling really hurt by their time at Nonlinear, and taking some months to recover before they could go back to work.”
“[Alice] reports a lot of effusive positive emotions from you, including various familial and sometimes romantic feelings of love, yet also felt threatened in various ways about her career and that she was in some pretty difficult financial circumstances as a result of working for you. These things seem pretty incongruent to me and potentially quite manipulative.”
“I’m not sure how to quickly summarize this. It seems to me they had a lot of strongly negative experiences or made decisions they regret that you (their employers and managers and the majority of their social environment) have some substantial responsibility for, in setting up this environment.”
“A number of people I spoke to were concerned about retaliation from you if they shared their experiences with me, and initially only did so under condition of strict confidentiality, which I was willing to offer because I had a bunch of warning flags raised from various people I trust who reached out to me and also from the financial situation Kat reported when we spoke at Lightcone.”
“[Alice] showed me some texts from Kat that offered support with “basic survival stuff” like housing, if Alice would “commit to not saying bad things about us to them”. This sort of behavior makes it very hard for me to trust impressions I get about your team — if people with negative information are being given strong reasons to keep it quiet, then I can’t find out about negative information about you, and I have a lot of uncertainty about how much I don’t know. A sense that people were scared to share negative info, and concerns about covering-up negative info, are the main reason I’m trying to find out what happened with [Alice] and [Chloe] and others as much as I am.
“[Alice] also showed me texts from Kat saying that [Alice] was saying bad things about you behind your back, and said that if you did the same to her her career in EA “would be over in a few Dm’s”. I’m interested to know what the content of those hypothetical DM’s would be. To be clear this definitely reads to me as a veiled threat.
I’m confused by the heavy disagree-voting on this comment. Are people saying with their votes that this does not actually cover all the important claims in the post? If someone who disagree-voted (or otherwise disagrees) wanted to comment I’d find that illuminating!
One of habryka’s other comments in this thread suggests that someone (or some small group of people, or some single person with sockpuppets) is downvoting literally everything habryka posts in this thread.
It would be nice if there were some way for someone with no conflict of interest to investigate that...
Where are you getting the idea that Ben gave us a week? The draft was sent to us on the same day Ben said he was going to publish it. On a day Ben knew we were traveling and wouldn’t be able to respond properly (sketchy/no internet, chaos of traveling, etc).
We spoke to Ben 60 hours before he published, and he only told us a subset of all the accusations. A quick re-reading of the post and I found 14 allegations that were new that Ben hadn’t discussed on the call. And I only got a short way through re-reading the post (maybe a 20%?) because I find reading it extremely painful.
I posted an approximately full list of claims you were informed of 4 days before publishing.
Would you be able to highlight any important claims that were not included in that list of claims? It is totally possible there are some, but having cross-check the two, I can’t find any major ones.
Were these fourteen included in the email Ben sent ~5d out (that Habryka summarizes here) and just not discussed on the call, or are you saying that the fourteen were first introduced to you with the final draft?
First introduced to me in the draft.
Thanks! If you were up for listing some of these fourteen I’d find it really helpful!
Strongly positive/negative relative to what? Relative to being more accurate initially, sure. Relative to being wrong but just not acknowledging it, no.
Specifically, I think people by default read this post as an example of good epistemic practice in a community built around good epistemic practice. In this case, though, I think the prior bad epistemic practice (not waiting for full information before publishing a highly consequential piece aimed at inflicting reputational damage on someone) is significant enough and bad enough that emphasizing a plan to update after the fact should be viewed primarily through the lens of damage control.
The standard with this sort of investigative piece should be to gather information of this nature prior to publication to whatever extent possible, where updating on new information after-the-fact is praiseworthy only if that information was not realistically knowable prior to publication.
You’re correct that it’s better than being wrong but not acknowledging it, but I think that’s a well-established standard in this sphere and there’s a stronger need to update based on the importance of gathering relevant information prior to publication.
Is anyone acting like that is not a damage control measure? I upvoted specifically because “do damage control” is better than “don’t”. Usually when I see a hit piece, and later there are a bunch of inaccuracies that come to light, I don’t in fact see that damage control done afterwards.
Also I think this kind of within-tribe conflict gets lots of attention within the EA and LW social sphere. I expect that if Ben publishes corrections a bunch of people will read them.
My read is that many people still consider the publication of the original post to be prudent and responsible given the circumstances, while any updates based on information that comes to light here will be prudent and responsible given the new information. Instead, I think people should view the original post as imprudent and irresponsible to the extent that it did not give one side of an adversarial situation an adequate hearing-out (and it really seems like it didn’t: a three-hour phone call where you misleadingly summarize their response as “Good summary!”, then refuse to wait until they can provide a more substantive response, is extraordinarily bad practice given the hundreds of hours he mentions putting into the rest of the investigation), with any subsequent updates being judged as returning towards responsibility after the fact rather than continuing a pattern of prudence.