FWIW, my model is also that the original post was received in a too witch-hunty manner, but also I don’t have any great ideas how to share the evidence to all the relevant parties without causing too much of a witch-hunt. I think all the specific statements that Ben made in his post were pretty well-calibrated (and still seem mostly right to me after reading through the evidence), so it’s not like the post called for a witch-hunt.
I’ve been thinking about whether there is some kind of informal court or arbitration system that would allow the social pressure here to be less driven by people trying to individually enact social enforcement, but by something that has more deliberation and moderation built-in, but I don’t yet have a blueprint for something that could work and also wouldn’t take thousands of hours.
If you have any concrete suggestions or edits for Ben’s post on what he could have done to make the effects be less witch-hunty, then I would be curious about that (though, to be clear, my overall assessment continues to be that working with Nonlinear is a bad idea, they should not have tables at EAG, should not receive central EA Funding, and young EAs should be reliable warned before engaging with them more, but like, not more than that. I don’t want people to try to actively harm Kat or Emerson, and I think it’s fine for them to work among themselves, build up an independent reputation and work on stuff they care about, and in as much as that happened, I am sad)
I’m surprised to hear you say this Habryka: “I think all the specific statements that Ben made in his post were pretty well-calibrated (and still seem mostly right to me after reading through the evidence)”
Do you think Ben was well calibrated/right when he made, for instance, these claims which Nonlinear has provided counter evidence for?
“She [Alice] was sick with covid in a foreign country, with only the three Nonlinear cofounders around, but nobody in the house was willing to go out and get her vegan food, so she barely ate for 2 days. Alice eventually gave in and ate non-vegan food in the house” (from my reading of the evidence this is not close to accurate, and I believe Ben had access to the counter evidence at the time when he published)
“Before she went on vacation, Kat requested that Alice bring a variety of illegal drugs across the border for her (some recreational, some for productivity). Alice argued that this would be dangerous for her personally, but Emerson and Kat reportedly argued that it is not dangerous at all and was “absolutely risk-free” (from my reading of the evidence Nonlinear provided, it seems Alice was asked to buy ADHD medicine that they believed was legal to buy where she was, and then they told her never mind when she said it required a prescription)
“After being hired and flying out, Chloe was informed that on a daily basis their job would involve driving e.g. to get groceries when they were in different countries.” (my understanding from what I read was that she was told she could take taxis paid for by nonlinear, and it was more like twice per week not daily)
“In summary Alice spent a lot of her last 2 months with less than €1000 in her bank account, sometimes having to phone Emerson for immediate transfers to be able to cover medical costs when she was visiting doctors. At the time of her quitting she had €700 in her account, which was not enough to cover her bills at the end of the month, and left her quite scared. ” (my understanding is that, according to nonlinear, this was not accurate)
“Afterwards, I wrote up a paraphrase of their responses. I shared it with Emerson and he replied that it was a “Good summary!”” [implying that Emerson was saying his characterization was accurate] (my understanding was that part of Emerson’s message was not mentioned, and that Emerson believed Ben’s summary had serious inaccuracies)
Do you think Ben was well calibrated/right when he made, for instance, these claims which Nonlinear has provided counter evidence for?
Yes, indeed I think in all of these quotes Ben basically said pretty reasonable things that still seem reasonably accurate to me even after reading the whole appendix that Nonlinear provided.
She [Alice] was sick with covid in a foreign country, with only the three Nonlinear cofounders around, but nobody in the house was willing to go out and get her vegan food, so she barely ate for 2 days. Alice eventually gave in and ate non-vegan food in the house”
You start with the one that I do think I made the biggest update on, though I also think most of the relevant evidence here was shared back during the original discussion. I am still kind of confused what happened here, and am hoping to dig into it, but I agree that there are some updates for me (and I assume others) here, and I currently think Alice’s summary is overall pretty misleading.
To be clear, in the quoted section Ben is summarizing what Alice told him, and Ben’s original post also directly includes this summary from Kat:
Second; the semi-employee said that she wasn’t supported in getting vegan food when she was sick with Covid, and this is why she stopped being vegan. This seems also straightforwardly inaccurate, we brought her potatoes, vegan burgers, and had vegan food in the house. We had been advising her to 80⁄20 being a vegan and this probably also weighed on her decision.
I think with both of these being listed, I am reasonably happy with the presentation and am glad that we included both sides on this.
You also say:
(from my reading of the evidence this is not close to accurate, and I believe Ben had access to the counter evidence at the time when he published)
This is inaccurate. I don’t think there is any evidence that Ben had access to that doesn’t seem well-summarized by the two sections above. We had a direct report from Alice, which is accurately summarized in the first quote above, and an attempted rebuttal from Kat, which is accurately summarized in the second quote above. We did not have any screenshots or additional evidence that didn’t make it into the post.
“Before she went on vacation, Kat requested that Alice bring a variety of illegal drugs across the border for her (some recreational, some for productivity). Alice argued that this would be dangerous for her personally, but Emerson and Kat reportedly argued that it is not dangerous at all and was “absolutely risk-free”
Yep, this still seems accurate to me. As Kat has documented herself she asked Alice to bring Schedule 2 drugs across borders without prescription (whether you need a prescription in the country you buy it is irrelevant, what matters is whether you have one in the country you arrive in), something that can have quite substantial legal consequences (I almost certainly would feel pretty uncomfortable asking my employee to bring prescription medications across borders without appropriate prescription).
I also currently still think that requests for other more recreational drugs were made, and I don’t consider the evidence to have debunked that, and wouldn’t have phrased the original very differently based on the evidence provided (I am hoping to find more concrete evidence for requests for recreational drugs, though it’s hard since as I understand these requests were verbal, if made at all).
Again, the original post also contains this summary of Kat’s position:
Third; the semi-employee was also asked to bring some productivity-related and recreational drugs over the border for us. In general we didn’t push hard on this. For one, this is an activity she already did (with other drugs). For two, we thought it didn’t need prescription in the country she was visiting, and when we found out otherwise, we dropped it. And for three, she used a bunch of our drugs herself, so it’s not fair to say that this request was made entirely selfishly. I think this just seems like an extension of the sorts of actions she’s generally open to.
This seems like a pretty good summary of what Nonlinear’s current position is, and so I feel pretty good about the details given here.
After being hired and flying out, Chloe was informed that on a daily basis their job would involve driving e.g. to get groceries when they were in different countries.
I don’t see any direct evidence for the daily vs. weekly claim, but I did update a bit on this dimension based on Kat directly claiming otherwise. My current best guess is that it was still multiple times a week, but not daily.
I do currently continue to believe that Chloe was pressured into driving without a license, and felt pressured to learn how to drive in the first place as part of her job. This kind of stuff is hard to arbitrate, and I don’t consider the evidence provided in the Nonlinear appendix here very compelling (it’s mostly just Kat asserting that she didn’t pressure her).
I currently feel a bit sad about the “daily” here, but would write roughly the same thing replacing it with “multiple times a week”.
In her comment, Chloe also explicitly says that getting taxis was hard, and often asking other people to drive her was the only option she saw, making me skeptical that the errands that she was expected to perform as part of her job were actually easily achievable via taxis:
And boy did we have to negotiate. I needed to run a medical errand for myself in Puerto Rico and the amount of negotiating I needed to do to get them to drive me to a different city that was a 30 min drive away was wild. I needed to go there three times, and I knew the answer of anyone driving me would be that it’s not worth their time, at the same time getting taxis was difficult while we were living in isolated mountain towns, and obviously it would have been easiest to have Drew or Emerson drive me
“In summary Alice spent a lot of her last 2 months with less than €1000 in her bank account, sometimes having to phone Emerson for immediate transfers to be able to cover medical costs when she was visiting doctors. At the time of her quitting she had €700 in her account, which was not enough to cover her bills at the end of the month, and left her quite scared. ”
I don’t currently see anything inaccurate in this paragraph.
I do think it’s currently sad that Ben’s post didn’t also mention Alice’s company, though my understanding is that she had made hard commitments to herself and her family to never withdraw any funds from that until it was properly self-sustaining, and had communicated those commitments to Kat and Emerson. It is still true that she did not have that money in her bank account in a direct way, and that she really didn’t want to withdraw that money for pretty reasonable reasons. I also think Nonlinear’s summary of Alice making $3000/mo from this company is inaccurate and off by a factor of 4x or 5x. My understanding is that Alice made less than $10k from this company throughout its whole existence (which is not very surprising, making money with that kind of company is very hard, especially if you are only spending a single day a week on it).
Nonlinear talks about the outstanding reimbursements, which Ben also directly included:
Though to be clear she was paid back ~€2900 of her outstanding salary by Nonlinear within a week, in part due to her strongly requesting it.
The “in part due to her strongly requesting it” part currently feels unclear to me, and I would probably say something slightly different given my current epistemic state. I genuinely believe that Alice was afraid and concerned that Nonlinear would withhold the €2900 from her, and it is clearly the case that she racked up $3000 worth of reimbursements on a $1000 dollar salary, which is pretty scary. The central point for this inclusion was to demonstrate the financial dependence. Saying instead “After she left she was paid back ~$3000 of outstanding reimbursements, which she was concerned Nonlinear might withhold from her or dispute being legitimate, which did help her financial situation” seems like it gets the same point across and captures my current epistemic state better.
I think it would have been better to say that she was paid back $3000 in outstanding reimbursements instead of salary, but I don’t think it makes a huge difference.
Afterwards, I wrote up a paraphrase of their responses. I shared it with Emerson and he replied that it was a “Good summary!”
Yeah, I think it was a bad move to not include the follow-up of “some points still require clarification”, but to be clear, I did totally interpret Emerson’s email as saying it was basically a good summary. I think it’s still bad to not have included the next sentence, and I was glad that it was corrected in the comments at the time (and I also somewhat think Ben should have updated the post to clarify that point so the correction is more easily visible).
For context, here is Emerson’s message:
Given the agreement-votes on this comment at the time, other people seemed to agree with me that summarizing the above as just a “Good Summary” was a kind of understandable mistake to make, and that this email did not successfully convey that Emerson disagreed with the points in the summary as provided,
Overall, yeah, given the messiness of a situation like this, I still feel quite good about basically all the quotes that you highlighted. I have now read the whole appendix to this post, spending over 10 hours going through it, and I can’t find any clear rebuttal to almost any of what I consider the core claims in Ben’s post.
Again, I do think someone might point out some inferences together with evidence in the appendix that might make me think some of the information that we quoted from sources is inaccurate. Both memory is fallible, and I also totally think that both Chloe and Alice had strong feelings here that might have clouded their judgement and memory. But as it stands, having spent a lot of time, I overall genuinely expected Nonlinear to be able to present more counter-evidence than Nonlinear was actually capable of providing, and currently think that any aggregate statements Ben made about probabilities he assigns to various behaviors, and the epistemic statuses he attached to statements made by Alice and Chloe, to still be quite well-calibrated and to capture the situation quite well.
It has still been less than a week, and the appendix is really massive, so if someone has clear rebuttals of things Ben said in the post, especially things that are not quotes or paraphrasings of Alice, but things that Ben directly claimed were true or where he expressed his epistemic state, I would be very interested in that. I have found a bit, but overall not very much.
You say: “This is inaccurate. I don’t think there is any evidence that Ben had access to that doesn’t seem well-summarized by the two sections above. We had a direct report from Alice, which is accurately summarized in the first quote above, and an attempted rebuttal from Kat, which is accurately summarized in the second quote above. We did not have any screenshots or additional evidence that didn’t make it into the post.”
Actually, you are mistaken, Ben did have screenshots. I think you just didn’t know that he had them. I can send you proof that he had them via DM if you like.
Regarding this: “As Kat has documented herself, she asked Alice to bring Schedule 2 drugs across borders without prescription (whether you need a prescription in the country you buy it is irrelevant, what matters is whether you have one in the country you arrive in), something that can have quite substantial legal consequences (I almost certainly would feel pretty uncomfortable asking my employee to bring prescription medications across borders without appropriate prescription).”
It sounds like you’re saying this paragraph by Ben:
“Before she went on vacation, Kat requested that Alice bring a variety of illegal drugs across the border for her (some recreational, some for productivity). Alice argued that this would be dangerous for her personally, but Emerson and Kat reportedly argued that it is not dangerous at all and was “absolutely risk-free”
is an accurate characterization of this sentiment: she was asked to pick up ADHD medicine in a place where it was believed not to require a prescription, and bring it back to a place where it does require a prescription, but then was told not to worry about it when it was found that it does require a prescription where she was going to pick it up.
To me, the former does a really bad job of capturing the latter, sounding WAY worse ethically. But I’d be curious to know if others agree with me or if they think that Ben’s paragraph captured this in a fair way.
On most of the other points I mentioned, it seems you feel that Ben made mistakes or ommisions, which I agree with:
“I currently feel a bit sad about the “daily” here, but would write roughly the same thing replacing it with “multiple times a week”.”
“I do think it’s currently sad that Ben’s post didn’t also mention Alice’s company”
“Yeah, I think it was a bad move to not include the follow-up of “some points still require clarification”″
On the taxi situation, I can’t speak to what it is like to get taxis in all the areas they lived, though this is one of those things Ben could have checked rather easily by asking where they were that she couldn’t get cabs and trying to book a cab (I don’t believe he bothered to check, but correct me if I’m wrong). When I stayed with all of the involved parties for a few days: Alice, Chloe, Kat, Emerson, and Drew (which was not in Peurto Rico), I got taxis twice, and it was slightly annoying—I spent about 5-10 minutes explaining where exactly to meet me, but was able to successfully get taxis on both of those occasions. Of course, that was just in one place that they lived (I think they were in that location for a couple of weeks if I recall correctly), and it might have been harder in other places, but at least in that location, getting taxis was no more than a minor nuisance.
Update: since it’s so easy to verify the claim about taxis, I just went ahead and checked it myself. My understanding is that Chloe was talking about Cabo Rojo, Puerto Rico. I used google maps to find and then call three taxi companies in that region. The first two didn’t pick up. The third said to text them, which I did, and they gave me a quote for getting a taxi today to drive 40 minutes (which was $45). It took about 10 minutes of my time (and 20 minutes on the clock, since they took ~10 minutes to give me a quote).
I didn’t realize this earlier, but in their evidence doc, Nonlinear talks about a similar check they did: “Lastly, it was not complicated to get a taxi there. I quickly checked, because sometimes we are in places that are truly remote. But there were three taxi services in the area that could have picked her up or driven her there. I called one of them and they said it would cost $30 and they could come pick me up whenever. And one of the other places where we were living at the time, she literally had to book a taxi for me out there, so I know she knew how and that it was possible.
There just wasn’t Uber there, so she’d have to make a phone call to a taxi.”
I can’t tell if you think Alice gave Ben basically accurate information and didn’t leave out critically important details, or if you think she did leave out critically important details (or directly lied to Ben), but it didn’t matter because Ben’s post was justified regardless.
Actually, you are mistaken, Ben did have screenshots. I think you just didn’t know that he had them. I can send you proof that he had them via DM if you like.
Sure! DMd you. I might also ping Ben, though want to mostly give him space and time to write a reply and not have to worry about stuff in the comments for now.
It sounds like you’re saying this paragraph by Ben:
“Before she went on vacation, Kat requested that Alice bring a variety of illegal drugs across the border for her (some recreational, some for productivity). Alice argued that this would be dangerous for her personally, but Emerson and Kat reportedly argued that it is not dangerous at all and was “absolutely risk-free”
is an accurate characterization of this sentiment: she was asked to pick up ADHD medicine in a place where it was believed not to require a prescription, and bring it back to a place where it does require a prescription, but then was told not to worry about it when it was found that it does require a prescription where she was going to pick it up.
To me, the former does a really bad job of capturing the latter, sounding WAY worse ethically. But I’d be curious to know if others agree with me or if they think that Ben’s paragraph captured this in a fair way.
On most of the other points I mentioned, it seems you feel that Ben made mistakes or ommisions, which I agree with:
No, that’s not what I am saying. I am saying that that paragraph, combined with Kat’s paragraph is a pretty reasonable summary of the evidence, and I think overall I think was pretty good at capturing the gist of the story (like, the two conflicting stories capture my present epistemic state still pretty well, which hasn’t changed very much).
My current model is that Kat is understating the degree to which she was asking Alice to carry drugs across borders, and so your summary (which is basically just a summary of things Kat has said) is also not an accurate summary of my epistemic state.
My current epistemic state is something like “Kat did ask Alice to bring over substantial number of prescription drugs, and probably also some recreational drugs that were illegal to bring across the border. My guess is most of the requests of this type were made in voice. I think the pressure here was substantial enough that I would feel very uncomfortable doing this to an employee of mine, and where if I saw the whole interaction I would think it’s a quite substantial flag, but also that no one did anything so egregious that this thing alone should cause any major repercussions for anyone involved”.
Agree that my epistemic state on this point is also something close to this.
Summarized would be “something like asking her to bring the drugs probably happened, and if so was a mistake that I’d hope was learned from, but the major issue would be if she was pressured to do it, and I’m unsure if I trust the person reporting enough to decide either way without evidence.”
[Edit: I know this is probably a frustrating thing for others to read, but seems worth saying anyway… since making the above comment I’ve had private information shared with me that makes me more confident NL didn’t act in an abusive way regarding this particular issue.]
Sure! DMd you. I might also ping Ben, though want to mostly give him space and time to write a reply and not have to worry about stuff in the comments for now.
Ok, I pinged Spencer. He sent me screenshots of text messages he sent Ben that he sent ~2 hours before publication of the post (in the middle of the barrage of comms that Nonlinear was firing off at the time, which included the libel threats), and which Kat posted to the comment thread less than 48 hours after the messages were sent to Ben.
I stand by my summary that everything Ben knew at the time of writing the post, made it into the post. Of course if you send something 2 hours before the post is published, late at night, it’s not going to make it into the post (but it might very well make it into a comment, which it did).
Ben made a bunch of other changes the day of publication. I know that because I pointed out errors in his post that day, and he was correcting them based on me pointing them out (e.g., all of his original quotes from glassdoor that he claimed were about Emerson were not actually about Emerson, which he didn’t realize until I pointed it out, and then he rushed to find new quotes to correct it). I’m sure he had a lot on his mind at that time, so I don’t think it’s egregious that he didn’t add mention of the fact that he had screen shot counter evidence about the “no food while sick stuff”, but it clearly seems to me to be a mistake on his part to not adjust the post or at least acknowledge it in the post. And I know he received the screenshots because he acknowledged getting them. You’re saying it made it into a comment as though Ben gets credit for that—but wasn’t it Kat who posted that comment? He also chose to rush the post out that night despite knowing there was counter evidence. I was honestly shocked he was trying to rush the post out that night because of all the errors I was finding in his post, which I expressed to Ben that day.
Update: I only just saw this point you made, including here for context and because it helps answer what I said: “since publishing this post required coordinating with many (5-10) external sources and witnesses, with many of them having a strong preference for a concrete time for the post to be published and they can plan around, so they can get ready for any potential retaliation and fallout. “
Update 2: “Spencer and Nonlinear knew the claims we were planning to put into the post on this matter roughly a week in-advance.” I’m not sure what you’re basing this on, but this is not accurate. I saw the post on Sept 6 for the first time and Ben published it early morning of Sept 7. During that short gap I sought to help him correct some errors in the post, which he did.
Update 3: the post was indeed being edited in meaningful ways on Sept 6, I know because I was helping Ben identify mistakes he had made and he was making changes based on that (such as the glass door misattribution)
Update 2: “Spencer and Nonlinear knew the claims we were planning to put into the post on this matter roughly a week in-advance.” I’m not sure what you’re basing this on, but this is not accurate. I saw the post on Sept 6 for the first time and Ben published it early morning of Sept 7. During that short gap I sought to help him correct some errors in the post, which he did.
Sorry, I had assumed that Nonlinear had shared with you the document we shared with them before publication of the post. My Slack records indicate Ben had two calls with you in the week before publication, and my guess based on the feedback I see from you paraphrased and copied in the Slack, is that you were aware of the claims in the post.
I might be wrong here and the Slack records could just align in a kind of confusing way (I can’t find the date of your first call with Ben, but am confident there was one more than 24 hours before publication), in which case I apologize.
Yes, Ben was making changes the day of the publication, I don’t think I said otherwise?
I also think sending something 2 hours before publication is again different from that (like clearly we can at least agree that if you had sent it 15 minutes before the publication time that it would not have been reasonable to say that Ben had access to information during the writing of the post that didn’t make it into the post?).
I really would not describe the post as being “rushed out”. The post had been worked on for over 1000 hours. I also think you are overstating “all the errors you were pointing out”. You pointed out two things which to me still seem relatively minor.
I think if Kat hadn’t posted the screenshots in a comment, Ben would have left a comment or edited the post. We really tried pretty hard to include anything that was sent to us, and I think Ben managed to include a lot of information and epistemic nuance in the post, while still maintaining the basics of readability and clarity.
When we did a postmortem on it, somewhat over 1000 hours is how high the total staff cost seemed to us, and that was a few months ago.
I think it’s totally plausible that in a few places I or someone else on the team used a lower number that they felt more confident in. In-general the structure of “over X” is something I usually use when I am not sure about X, but want to give a quick lower bound that allows me to move ahead with the argument, so it seems totally possible that in another context I would have said “multiple hundreds of hours” or “300+” hours or something like that, because that was enough to prove the point at hand.
Edit: Oh, I see the links now, didn’t see them when I first wrote the comment.
I kept wanting to just share what I’d learned. I ended up spending about ~320 hours (two months of work), over the span of six calendar months, to get to a place where I was personally confident of the basic dynamics (even though I expect I have some of the details wrong), and that Alice and Chloe felt comfortable with my publishing.
I think the key difference with that quote and my number is that it just includes Ben’s time, as opposed to total staff time. For example, it omits work done by anyone else on the team (which roughly doubles the total amount of time spent, spread across me, Robert and Ruby), as well as others who we’ve brought on board to help with the post (we worked with 2-3 external collaborators who ended up pairing with Ben for multiple weeks).
My guess is also Ben’s number is a bit low for his own time spent on it, though I think we are now getting into definitions of what counts as “working on it”. We don’t have detailed time tracking, so this is a bit hard to operationalize, but my guess is if you added up all the staff time of Lightcone staff and external collaborators, and removed the project of writing the Nonlinear post, you would indeed end up with somewhat more than a thousand hours of additional free time across those people.
Thanks. I didn’t mean my comment to come across as a “gotcha” question fwiw (not saying that you said it was a gotcha question, but I realized after I commented that it’d be a reasonable interpretation of my comment).
For what it’s worth, I find it extremely plausible that a post like this both took an inordinately large amount of time, and that people will systematically underestimate how much it took before they started doing more accurate time-tracking.
It does seem very sad that the voting on this post seems a bit broken (it also seemed broken on the original Nonlinear post). Like, do people think I am lying about the amount of hours it took? I would be happy to provide the data that I have, or have someone else who is more independent to the Lightcone team provide an estimate. It seems very weird to downvote an answer to a straightforward question like that.
Hmm, well Ben said “(for me) a 100-200 hour investigation” in the first post, then said he spent “~320 hours” in the second. Maybe people thought you should’ve addressed that discrepancy?️ Edit: the alternative―some don’t like your broader stance and are clicking disagree on everything. Speaking of which, I wonder if you updated based on Spencer’s points?
Edit: Oh, I see the links now, didn’t see them when I first wrote the comment.
Apologies, that was my fault. I wrote the comment and then I realized that I was demonstrating poor reasoning transparency, so then I hunted down the relevant links. My guess of chronology was that I had the hyperlinks added in after you started commenting, but before your reply was visible.
Ah, cool, I was really surprised when I saw the links on refresh, but they fit so naturally into the comment that I thought they clearly must have been there in the first place.
if you send something 2 hours before the post is published, late at night, it’s not going to make it into the post.
This would make sense to me if Ben had been working to an external deadline, but instead this is directly downstream from Ben’s choice to allocate very little time to draft review and ensuring he had his facts right. It sounds like Spencer sent these text messages <24hr after being sent the draft; how quickly would he have needed to turn around his review to count?
To be clear, we were working to a substantial degree to an external deadline, since publishing this post required coordinating with many (5-10) external sources and witnesses, with many of them having a strong preference for a concrete time for the post to be published and they can plan around, so they can get ready for any potential retaliation and fallout.
There was wiggle room in that date and time, but by the time Spencer sent this, the post and publish-date was really quite locked in.
I think 24 hours before publication would have been enough to include them. Maybe even 12 hours. As I mentioned in other places, we did send Nonlinear (and Spencer) a list of the relevant claims in the final post, including this one, so I think the fact that the literal draft was only shared 24 hours in advance is irrelevant. Spencer and Nonlinear knew the claims we were planning to put into the post on this matter roughly a week in-advance.
For example, in the call that Nonlinear cancelled with us a day before publication, that would have been a pretty good time to share such evidence with us, and if they had given additional evidence then, it would have made it into the post.
But separately from that, I am not sure what you mean by “count”. Spencer claimed that “we had screenshots that didn’t make it into the post”. I think a reasonable reader would infer from that when the post was being written, we had access to those screenshots. By the time Spencer sent these screenshots, the post was no longer being written in any meaningful way.
To be clear, we were working to a substantial degree to an external deadline, since publishing this post required coordinating with many (5-10) external sources and witnesses, with many of them having a strong preference for a concrete time for the post to be published and they can plan around, so they can get ready for any potential retaliation and fallout.
I don’t really see how this is a defense. The fact that you have promised some third parties to do X does not justify you in doing X if X would otherwise not be morally acceptable. And publishing harmful statements about someone that you have good reason to think are false does not seem morally acceptable.
It is a defense that in as much as I think anyone working on a post similar to this, mostly independently of skill level, would end up having to make promises to sources of this type, in order to be able to share concerning information publicly.
Of course, if you think posts of this whole reference class are bad, and it was bad for us to even attempt to make a post that tries to publicize the extensive rumors and concerns that we heard about Nonlinear, then I think it’s not a defense.
But if you think people should attempt to spread that kind of information and share it with more parties, then I think this will somewhat inevitable come with constraints like having to keep publication deadlines and coordinating the many stakeholders involved in such a thing.
Like, what is the alternative that you propose we should have done instead? Not made any promises to our sources at all about doing things that protect them from retaliation and limiting the costs on them? I think in that case you don’t get to talk to sources, or you only get to do it for a bit as people get burned and hurt and stop talking to you.
And publishing harmful statements about someone that you have good reason to think are false does not seem morally acceptable.
I am pretty sure Ben has published no harmful statements about someone that he thought were false. Indeed, as I have said many times, he seems to have been exceptionally careful with the epistemic states he attached to his statements in his post.
I’m well aware of the difficulties of balancing competing stakeholders giving you feedback late on posts and trying to hit publication timing targets. I think you had several valid options:
Never make commitments about publication date and time in the first place.
Make commitments, but be clear they are provisional. When you receive this information, email your sources saying “hey guys, really sorry but we just received some last-minute info that we need to update on. We’ll circle back to coordinate a new launch date that works for you.”
Give Spencer a reasonable deadline to respond, committing to take into account feedback received before this deadline.
Delete that section and publish on the original schedule.
Edit the section and publish on the original schedule.
Edit the section and publish on the original schedule.
I mean, to be clear, we did this the first time Nonlinear disputed the relevant section.
Alice quit being vegan while working there. She was sick with covid in a foreign country, with only the three Nonlinear cofounders around, but nobody in the house was willing to go out and get her vegan food, so she barely ate for 2 days. Alice eventually gave in and ate non-vegan food in the house. She also said that the Nonlinear cofounders marked her quitting veganism as a ‘win’, as they thad been arguing that she should not be vegan.
(Nonlinear disputes this, and says that they did go out and buy her some vegan burgers food and had some vegan food in the house. They agree that she quit being vegan at this time, and say it was because being vegan was unusually hard due to being in Puerto Rico. Alice disputes that she received any vegan burgers.)
I think this section is really quite clear. We have one report from Alice saying that she quit being vegan. We directly include, in the next paragraph, the fact that Nonlinear disputes this. I really don’t think we misled anyone. The screenshots sent did not actually materially change anything in the paragraphs above, indeed both of the paragraphs are still fully accurate (and in as much as Alice claimed that she did not get food while indeed getting food, that is IMO an important part of the story that seems important for other people to be able to cross-check).
I think the choice of “you have some sources, you cite the sources while being really quite clear that you don’t fully trust your sources, and when a thing gets directly disputed by another source you say that directly” is a reasonable thing to do. Again, as I’ve said an enormous number of times, we never had an intention of fully litigating all of these claims before publication, which would have been completely infeasible time-wise.
The alternative to Ben’s post would have probably been a series of fully anonymous posts with extremely vague high-level accusations that would have been extremely hard to respond to. We tried to make the claims concrete and provide an interface to aggregate information at all.
Like, what kind of edit would you have preferred us to do instead of the above?
I think this section is really quite clear. We have one report from Alice saying that she quit being vegan. We directly include, in the next paragraph, the fact that Nonlinear disputes this. I really don’t think we misled anyone.
I strongly disagree. Alice’s and Nonlinear’s perspectives are portrayed with very different implicit levels of confidence in those paragraphs. Alice’s perspective is stated as a fact—“nobody in the house was willing to go out and get her vegan food,” not “Alice says nobody in the house was willing to go out and get her vegan food.” In contrast, Nonlinear’s perspective is shared as “[Nonlinear] says [x].”
I think most readers who trust Ben to be truthful would assume, from the way those paragraphs were worded, that Alice had much better evidence to support her claims, and that Nonlinear was doing some slightly deceitful reputational management by countering them. But that isn’t what turned out to be the case:
Nonlinear has evidence that on December 15, they had oatmeal, peanuts, almonds, prunes, tomatoes, cereal, an orange, mixed nuts, and quinoa (which Kat offered to cook) in the house.
At some point, Emerson went out and tried to purchase Alice more food despite his knee injury, but he couldn’t find the very specific items she requested.
Then, on December 18, it looks like Alice’s first non-vegan meal was a vegetarian pizza she ordered (rather than non-vegan food already in the house). It looks like she ordered it right after Kat reminded her that they already had vegan noodles in the fridge.
On top of all of this, apparently everybody in the house was either sick or injured, but Ben’s post only mentions that Alice was sick.
It seems that Alice/Ben have no evidence to counter any of the points above.
So the original claim that was stated as fact (“nobody in the house was willing to go out and get her vegan food”) seems very wrong. Which is sad, because it’s a very serious accusation that most people would assume was not made lightly.
Yeah, sorry, I think I was too strong in my language above, though my sense is you are also interpreting my answer to be about a somewhat different question than the one I perceived Larks as asking. To clarify where I think we are on the same page: I am pretty unhappy about that section, and wouldn’t ask Ben to write something different given what I believe today.
The thing I was responding to was whether we misrepresented the evidence that we did have at the time.
On that topic, I do think it was a mistake to omit as many of the “Alice/Chloe claims that X” in the post as we did, and fall back into a neutral third-party way of summarizing the claims, and given that we did, I think it makes sense to hold Ben and Lightcone more responsible for the veracity of statements that did not include an explicit “Alice/Chloe alleges X”.
I also think that there is a pretty reasonable case to be made that we should have waited longer on getting more evidence from Nonlinear. I felt conflicted on this topic then, and feel conflicted now. I really hate that the situation we were in made it quite hard for us to wait longer for Nonlinear to respond to us. I am still not fully sure whether I would wait if I was in this situation again, since the considerations against waiting were also quite strong, though overall I am leaning slightly that waiting would have been the better option (I do not think this forgives or excuses Nonlinear’s attempts at intimidation and threats of retaliation).
However, overall on the question of “did we accurately summarize the evidence available to us”, I think Ben’s post and this section is doing pretty well.
I agree that we frame Alice and Chloe’s evidence as more trustworthy, and in-aggregate, across the whole post, I stand behind that framing, in that I think Alice and Chloe are substantially more reliable sources of evidence than Kat and Emerson. I agree that in this situation I think this went the wrong way around and it looks to me like the vegan food situation seems like it was represented to us in a substantially misleading way, and I am still hoping for me or Ben to follow up with Alice on this topic and figure out whether I am missing something. However, I think on-average the framing of the post was not misleading about the balance of evidence that we had received to that point (including accounting for expected future evidence Nonlinear that we expected Nonlinear might provide).
Some smaller nitpicks on your comment:
In contrast, Nonlinear’s perspective is shared as “[Nonlinear] says [x].”
It’s true that we don’t share Nonlinear’s perspective with the same authority as Alice and Chloe’s. We did also include a summary directly written from their perspective, which I do think helps:
This seems also straightforwardly inaccurate, we brought her potatoes, vegan burgers, and had vegan food in the house. We had been advising her to 80⁄20 being a vegan and this probably also weighed on her decision.
Another quick comment:
It seems that Alice/Ben have no evidence to counter any of the points above.
I would give people some time before concluding that. While Ben (and I) are trying really hard to not be dragged into a full-on follow-up investigation of this, I do expect there will be some kind of response to this which includes procuring more evidence. I personally do feel pretty convinced on this point, but I am not updating on Alice or Ben not providing more evidence in coming to that conclusion, since they haven’t responded to anything so far, and I do know that many of the claims in the OP and associated appendix are inaccurate, and those also haven’t been responded to yet (it includes many claims about what Ben believes or what the process of writing Ben’s original post was like, which I am very confident are inaccurate).
I am substantially less confident in that claim, though yeah, I would still overall say I believe it (it’s not super well-operationalized so not super clear what a probability would mean, but like, I guess I am at ~80% that if I knew all the facts and had arbitrary insight into Alice’s, Kat’s and Emersons’ life that I would overall expect Alice to be reporting more accurately than Kat and Emerson)
I’m not sure if Spencer sent you all the screenshots or just some of them, but something along the lines of:
Alice quit being vegan while working there. She was sick with covid in a foreign country, with only the three Nonlinear cofounders around, who she alleges refused to go out and get her vegan food, so she barely ate for 2 days. Alice eventually gave in and ate non-vegan food in the house. She also said that the Nonlinear cofounders marked her quitting veganism as a ‘win’, as they thad been arguing that she should not be vegan.
(Nonlinear disputes this, and sent dated screenshots suggesting and says that they did go out and buy her some vegan burgers food and had some vegan food in the house that they cooked for her. They agree that she quit being vegan at this time, and say it was because being vegan was unusually hard due to being in Puerto Rico. Alice disputes that she received any vegan burgers; we did not ask her to comment on the screenshots.)
Importantly the screenshot only covered events on December 15th.
Here is the relevant screenshot:
Kelsey Piper in the thread summarizes these screenshots (together with some other screenshots that Kat shared) as follows:
On December 15, Alice states that she’d had very little to eat all day, that she’d repeatedly tried and failed to find a way to order takeout to their location, and tries to ask that people go to Burger King and get her an Impossible Burger which in the linked screenshots they decline to do because they don’t want to get fast food. She asks again about Burger King and is told it’s inconvenient to get there. Instead, they go to a different restaurant and offer to get her something from the restaurant they went to. Alice looks at the menu online and sees that there are no vegan options. Drew confirms that ‘they have some salads’ but nothing else for her. She assures him that it’s fine to not get her anything.
It seems completely reasonable that Alice remembers this as ‘she was barely eating, and no one in the house was willing to go out and get her nonvegan foods’ - after all, the end result of all of those message exchanges was no food being obtained for Alice and her requests for Burger King being repeatedly deflected with ‘we are down to get anything that isn’t fast food’ and ‘we are down to go anywhere within a 12 min drive’ and ‘our only criteria is decent vibe + not fast food’, after which she fails to find a restaurant meeting those (I note, kind of restrictive if not in a highly dense area) criteria and they go somewhere without vegan options and don’t get her anything to eat.
It also seems totally reasonable that no one at Nonlinear understood there was a problem. Alice’s language throughout emphasizes how she’ll be fine, it’s no big deal, she’s so grateful that they tried (even though they failed and she didn’t get any food out of the 12⁄15 trip, if I understand correctly). I do not think that these exchanges depict the people at Nonlinear as being cruel, insane, or unusual as people. But it doesn’t seem to me that Alice is lying to have experienced this as ‘she had covid, was barely eating, told people she was barely eating, and they declined to pick up Burger King for her because they didn’t want to go to a fast food restaurant, and instead gave her very limiting criteria and went somewhere that didn’t have any options she could eat’.
My guess is this aligned with Ben’s interpretation at the time. The screenshots were relevant evidence, but they did not directly disprove anything in the original article.
Kat then shared further screenshots in the comments, which importantly were not shared with Ben beforehand(unless Spencer failed to forward them to me in my DM with him yesterday), that demonstrated that on the next day Kat did successfully bring her food.
However, the story, in the above screenshot, on December 15th, is that indeed Alice did not get food, despite her requesting it. The screenshots that Spencer sent us appear to fail to include the most relevant part of the conversation, which is that they did indeed fail to get her vegan food that day during that trip.
(Edit: Kat disputes this below, sharing some additional screenshots that seem to show that Kat did get food for Alice later that day, which seems important to get right. Though I don’t think Spencer’s screenshot demonstrated this).
Here are the edits I currently agree would have been better, though I think they are minor enough that I don’t currently see it as a major error to not have included them:
Nonlinear disputes this, and sent dated screenshotsthat document part of the relevant conversation. In those screenshots we can see that on December 15th Alice did indeed request vegan food from multiple restaurants, but her and Drew ran into difficulties finding vegan food that was available in the area, and it seems like Nonlinear ultimate did indeed decline to stop by a different restaurant for Alice to get her vegan food, though the details are not fully clear. They agree that she quit being vegan at this time, and say it was because being vegan was unusually hard due to being in Puerto Rico. Alice disputes that she received any vegan burgers; we did not ask her to comment on the screenshots.)
I really encourage you to look at the screenshots, Kelsey’s summary, and Kat’s original comment on the Nonlinear post and explain to me how these screenshots falsify part of the post. As we later received more screenshots, it seems like we actually received confirmation that the conversation on that date did indeed not result in Alice getting food.
(Edit: Kat shares some additional screenshots below that do seem to show Alice got food on the 15th, though not from the restaurant trip that was talked about in the screenshot Spencer sent us)
I’m a little bit confused about Kelsey’s summary—it contains a line about rejecting burgers because they were ‘fast food’ that doesn’t seem to be in the original. So I don’t think it can reflect Ben’s state of mind in that way.
If you only had the one screenshot (9:53 to 10:28 timestamps), I agree that you can’t infer that Kat cooked for ‘Alice’, nor is there proof that the discussed burger trip actually took place, though I think they strongly imply it will—certainly Alice seems to think it has been agreed and will occur. However, I find your comment about 15th vs 16th unconvincing because ‘Alice’ explicitly claims a 2 day duration, so food the next day would also contradict this (assuming the 15th is the first day).
Here is another possible version that reflects just the one screenshot:
Alice quit being vegan while working there. She was sick with covid in a foreign country, with only the three Nonlinear cofounders around, who she alleges refused to go out and get her vegan food, so she barely ate for 2 days. Alice eventually gave in and ate non-vegan food in the house. She also said that the Nonlinear cofounders marked her quitting veganism as a ‘win’, as they thad been arguing that she should not be vegan.
(Nonlinear disputes this, and sent dated screenshots showing a conversation Kat had with ‘Alice’ about two different restaurants they could go to, as well as the vegan food they had in the house, which seems to end with an agreement to go to get (vegan) burgers. They further claim (though the screenshots do not prove this) and says that they did go out and buy her some vegan burgers food and had some vegan food in the house. They agree that she quit being vegan at this time, and say it was because being vegan was unusually hard due to being in Puerto Rico. Alice disputes that she received any vegan burgers; we did not ask her to comment on the screenshots.)
I think my key objections to the original version is asserts as fact that Nonlinear refused to get her food, vs this being an unverified claim, and that it does not reflect that NonLinear didn’t merely dispute it, they offered evidence.
If you only had the one screenshot (9:53 to 10:28 timestamps), I agree that you can’t infer that Kat cooked for ‘Alice’, nor is there proof that the discussed burger trip actually took place, though I think they strongly imply it will
Just to be clear, that burger trip did indeed not happen that day, if I understand it correctly. What instead happened is that Kat went out a few hours later and got Alice mashed potatoes at a store (which is not really hinted at at all in the screenshots).
Yeah, I think this version is reasonable and I would have preferred to post this version (and somewhat think that we should have updated it ASAP, even after publication).
on December 15th, is that indeed Alice did not get food,
This is false. Alice got food on December 15th. She got food 2.5 hours after she asked. Actually, she never asked me, I just offered when it seemed like she was struggling.
It says December 16 at 12:14am because I was in Europe at the time, so it’s showing the European time zone. It was Dec 15 at 7:13pm in the local time when this occurred.
She brought up being hungry at 4:53pm. I immediately offered to cook her the food in the house. When she didn’t want any of the food in the house or food from any non-fast food restaurant within a 12 minute drive of home, I went out, while sick myself, and got and cooked her food. The only vegan food that fit her criteria in the store.
The only complaint she can legitimately say is that we did not get her Panda Express as fast as she would have liked (we got it for her the next day). She waited 2.5 hours for food. And she could have had it sooner if she’d wanted any of the food in the house, which she usually ate nearly daily and enjoyed. She just didn’t want that food. She wanted fast food and didn’t get it as fast as she preferred.
Thank you! This definitely seems like highly relevant evidence.
Can you clarify whether Kelsey’s summary of the December 15th conversation is accurate or inaccurate? It’s totally possible that I am misreading the screenshots, though my best interpretation was indeed the interpretation that Kelsey made in the screenshots.
I would be happy to correct the statement above if I am wrong here.
I do think this issue seems somewhat separate from the question of “did the screenshots that were shared with us materially affect the things Ben wrote?”.
To be clear, this is relevant in as much as the original screenshot was evidence of there being more things you could share here, though I currently maintain that I don’t think the screenshots that were shared with us showed any material error (given that Kelsey also walked away with the same impression of them being consistent).
I also totally care about just setting the record straight and getting the object-level issue right here, and in as much as there isn’t anything very weird going on with the screenshots you sent, I think you provided pretty decent proof here and am changing my mind on the December 15th issue (and think if you had shared those screenshots with us instead, I think it’s pretty likely Ben would have somehow made sure that they made it into the post).
Kelsey’s summary was wrong in a number of important ways.
She missed the fact that we did indeed succeed in getting her vegan food (I found at the nearby store, despite being sick myself). 2.5 hours after we first offered. And it would have been faster if she’d wanted any of the food in the house, or chosen a restaurant that had vegan options for Emerson and Drew to go to.
It doesn’t mention the vegan food that was in the house already that I offered to cook (Alice ate oatmeal almost every day and she loved quinoa. Later when I cooked some up for her, she loved it, like usual, cause quinoa is the Queen of All Foods).
It doesn’t mention that Drew said he would go to any restaurant within a 12 minute drive from our place and she just… didn’t choose a restaurant. She only wanted fast food. So they ended up choosing a restaurant that didn’t happen to have vegan options aside from the usual fries.
A quick look at Google Maps shows that there was over 20 restaurants that fit that criteria in the area. It wasn’t restrictive at all.
She frames it as they didn’t get her the food she wanted “because they [didn’t] want to get fast food.” It’s important to note that Emerson and Drew also are people whose preferences matter. Just because Alice is sick doesn’t mean everybody has to drop their own needs and preferences to get her the very particular food she wants.
She frames it as Emerson and Drew being somehow inconsiderate and shallow, when you could just as easily frame it as Alice not considering the needs or preferences of anybody but herself, expecting everybody to drop everything and go out of their way so she can get the very specific fast food she wants. Then, when she doesn’t get exactly what she wants as fast as she wants, she goes around telling lies about what happened to destroy a charity (e.g. nobody willing to get her food)
As for what Ben knew before publishing, if you look at that screenshot, you can see that:
Drew has offered to pick her up food (“Drew suggested he could otherwise pick up stuff”)
Me and Emerson offered to pick her up food (“Me and Emerson can do it if he can’t”)
I offered to cook her the food in the house (“Could make you some quinoa”)
Ben said in his post that “nobody in the house was willing to go out and get her vegan food”. This is absolutely false. Ben had seen this screenshot clearly showing that we were willing to go out and get her food.
Note that she didn’t even ask for food. I just offered because I could see she was in need (“want me to order food?”)
It also said “so she barely ate for 2 days”. It shows in the messages that there was vegan food in the house. So that was clearly not the reason she didn’t eat for days.
She had plenty of options but she wanted fast food in particular.
We got her her first choice of fast food the very next day (remember, she only started asking for food in the evening the previous day. It was also hard to get stuff for her. This was our first experience with covid and we were trying to figure out how to manage it, try to have it not spread, etc etc. It was quite a stressful and overwhelming time.).
Because she didn’t get a very particular fast food as fast as she wanted, she interpreted this as us being heartless people who wouldn’t take care of a sick person in need. She told Ben a false and misleading story about us not being willing to go out and get food for her.
As we later received more screenshots, it seems like we actually received definitive confirmation that the conversation on that date did indeed not result in Alice getting food.
I’m waiting for Ben, or someone else, to make a table of claims, counter claims, and what the evidence shows. Because nonlinear providing evidence that doesn’t support their claims seems to be a common occurance.
Just to give a new example, Kat screenshots herselfreplying “mediating! Appreciate people not talking to loud on the way back [...] ” here, to provide evidence supporting that there was not a substantial discussion that occurred. However, I can only interpret the use of “mediating!” to indicate that there was in-fact a substantial amount of discussion at play.
Edit: Retracted as correctly pointed out by @Sean_o_h , I read meditation as mediation.
Ben’s choice to allocate very little time to draft review and ensuring he had his facts right
Can you clarify what you mean by “very little time”? Haybrka reports spending 1000+ staff hours, and even Ben’s much more conservative estimate of 100-200 hours doesn’t feel fair to me to describe as “very little”
Sorry, I’m trying to talk about the amount of time for ‘adversarial’ fact checking: when Nonlinear knew the accusations and could provide specific counter evidence. I agree he put a ton of time into the project overall.
Just a note that standard practice on these kinds of jobs is that you get a credit card to make purchases with, and are never using your own money that is later reimbursed.
A big reason for this is the massive mismatch in what money is worth. Employers might think covering a $100 grocery trip until you get reimbursed is not a big deal, but to an employee that might have been their own food money or rent.
The standard answer is you either let your employee borrow your credit card, or you give them their own credit card. You can put a lower limit on it to protect yourself, and can also see the credit card statement (which can be paired with receipts if you don’t trust them not to add on extras. I was always careful that my families get all the receipts but they generally just threw them away because they trusted me)
I’ve been thinking about whether there is some kind of informal court or arbitration system that would allow the social pressure here to be less driven by people trying to individually enact social enforcement
My model has been there should be social enforcement for both poor epistemic practices and rude/unkind communication.
I have been an active commenter in both posts, with a goal of social pressure in mind (i.e. providing accountability and a social pressure to not behave inappropriately towards/with your employees).
I’d be interested to hear meta level criticisms of my approach (e.g. “social pressure is inherently bad”). Because, whilst I don’t want witch hunting that employs poor epistemic practices, I do think social pressure plays an important role in stabilising communities. Perhaps someone can change my mind on this? If you do change my mind, I’ll certainly comment a lot less.
To me it seems like everyone individually applying social pressure is hard to calibrate. Oli seems to be saying that he and Ben did not intend the level of social consequences NL has felt based on what they shared, but rather an update that NL shoudn’t be a trusted EA org. I think that it’s hard to control the impression that people will get when you provide a lot of evidence even if it’s all relatively minor, and almost impossible to control snowballing dynamics in comment sections and on social media when people fear being judged for the wrong reaction, so it just might not be possible for a post like Ben’s to received in a calibrated way.
This sounds right, but the counterfactual (no social accountability) seems worse to me, so I am operating on the assumption it’s a necessary evil.
I live high trust country, which has very little of this social accountability, i.e. if someone does something potentially rude or unacceptable in public, they are given the benefit of the doubt. However, I expect this works because others are employed, full time, to hold people accountable. I.e. police officers, ticket inspectors, traffic wardens. I don’t think we have this in the wider Effective Altruism community right now.
FWIW, my model is also that the original post was received in a too witch-hunty manner, but also I don’t have any great ideas how to share the evidence to all the relevant parties without causing too much of a witch-hunt. I think all the specific statements that Ben made in his post were pretty well-calibrated (and still seem mostly right to me after reading through the evidence), so it’s not like the post called for a witch-hunt.
I’ve been thinking about whether there is some kind of informal court or arbitration system that would allow the social pressure here to be less driven by people trying to individually enact social enforcement, but by something that has more deliberation and moderation built-in, but I don’t yet have a blueprint for something that could work and also wouldn’t take thousands of hours.
If you have any concrete suggestions or edits for Ben’s post on what he could have done to make the effects be less witch-hunty, then I would be curious about that (though, to be clear, my overall assessment continues to be that working with Nonlinear is a bad idea, they should not have tables at EAG, should not receive central EA Funding, and young EAs should be reliable warned before engaging with them more, but like, not more than that. I don’t want people to try to actively harm Kat or Emerson, and I think it’s fine for them to work among themselves, build up an independent reputation and work on stuff they care about, and in as much as that happened, I am sad)
I’m surprised to hear you say this Habryka: “I think all the specific statements that Ben made in his post were pretty well-calibrated (and still seem mostly right to me after reading through the evidence)”
Do you think Ben was well calibrated/right when he made, for instance, these claims which Nonlinear has provided counter evidence for?
“She [Alice] was sick with covid in a foreign country, with only the three Nonlinear cofounders around, but nobody in the house was willing to go out and get her vegan food, so she barely ate for 2 days. Alice eventually gave in and ate non-vegan food in the house” (from my reading of the evidence this is not close to accurate, and I believe Ben had access to the counter evidence at the time when he published)
“Before she went on vacation, Kat requested that Alice bring a variety of illegal drugs across the border for her (some recreational, some for productivity). Alice argued that this would be dangerous for her personally, but Emerson and Kat reportedly argued that it is not dangerous at all and was “absolutely risk-free” (from my reading of the evidence Nonlinear provided, it seems Alice was asked to buy ADHD medicine that they believed was legal to buy where she was, and then they told her never mind when she said it required a prescription)
“After being hired and flying out, Chloe was informed that on a daily basis their job would involve driving e.g. to get groceries when they were in different countries.” (my understanding from what I read was that she was told she could take taxis paid for by nonlinear, and it was more like twice per week not daily)
“In summary Alice spent a lot of her last 2 months with less than €1000 in her bank account, sometimes having to phone Emerson for immediate transfers to be able to cover medical costs when she was visiting doctors. At the time of her quitting she had €700 in her account, which was not enough to cover her bills at the end of the month, and left her quite scared. ” (my understanding is that, according to nonlinear, this was not accurate)
“Afterwards, I wrote up a paraphrase of their responses. I shared it with Emerson and he replied that it was a “Good summary!”” [implying that Emerson was saying his characterization was accurate] (my understanding was that part of Emerson’s message was not mentioned, and that Emerson believed Ben’s summary had serious inaccuracies)
Yes, indeed I think in all of these quotes Ben basically said pretty reasonable things that still seem reasonably accurate to me even after reading the whole appendix that Nonlinear provided.
You start with the one that I do think I made the biggest update on, though I also think most of the relevant evidence here was shared back during the original discussion. I am still kind of confused what happened here, and am hoping to dig into it, but I agree that there are some updates for me (and I assume others) here, and I currently think Alice’s summary is overall pretty misleading.
To be clear, in the quoted section Ben is summarizing what Alice told him, and Ben’s original post also directly includes this summary from Kat:
I think with both of these being listed, I am reasonably happy with the presentation and am glad that we included both sides on this.
You also say:
This is inaccurate. I don’t think there is any evidence that Ben had access to that doesn’t seem well-summarized by the two sections above. We had a direct report from Alice, which is accurately summarized in the first quote above, and an attempted rebuttal from Kat, which is accurately summarized in the second quote above. We did not have any screenshots or additional evidence that didn’t make it into the post.
Yep, this still seems accurate to me. As Kat has documented herself she asked Alice to bring Schedule 2 drugs across borders without prescription (whether you need a prescription in the country you buy it is irrelevant, what matters is whether you have one in the country you arrive in), something that can have quite substantial legal consequences (I almost certainly would feel pretty uncomfortable asking my employee to bring prescription medications across borders without appropriate prescription).
I also currently still think that requests for other more recreational drugs were made, and I don’t consider the evidence to have debunked that, and wouldn’t have phrased the original very differently based on the evidence provided (I am hoping to find more concrete evidence for requests for recreational drugs, though it’s hard since as I understand these requests were verbal, if made at all).
Again, the original post also contains this summary of Kat’s position:
This seems like a pretty good summary of what Nonlinear’s current position is, and so I feel pretty good about the details given here.
I don’t see any direct evidence for the daily vs. weekly claim, but I did update a bit on this dimension based on Kat directly claiming otherwise. My current best guess is that it was still multiple times a week, but not daily.
I do currently continue to believe that Chloe was pressured into driving without a license, and felt pressured to learn how to drive in the first place as part of her job. This kind of stuff is hard to arbitrate, and I don’t consider the evidence provided in the Nonlinear appendix here very compelling (it’s mostly just Kat asserting that she didn’t pressure her).
I currently feel a bit sad about the “daily” here, but would write roughly the same thing replacing it with “multiple times a week”.
In her comment, Chloe also explicitly says that getting taxis was hard, and often asking other people to drive her was the only option she saw, making me skeptical that the errands that she was expected to perform as part of her job were actually easily achievable via taxis:
I don’t currently see anything inaccurate in this paragraph.
I do think it’s currently sad that Ben’s post didn’t also mention Alice’s company, though my understanding is that she had made hard commitments to herself and her family to never withdraw any funds from that until it was properly self-sustaining, and had communicated those commitments to Kat and Emerson. It is still true that she did not have that money in her bank account in a direct way, and that she really didn’t want to withdraw that money for pretty reasonable reasons. I also think Nonlinear’s summary of Alice making $3000/mo from this company is inaccurate and off by a factor of 4x or 5x. My understanding is that Alice made less than $10k from this company throughout its whole existence (which is not very surprising, making money with that kind of company is very hard, especially if you are only spending a single day a week on it).
Nonlinear talks about the outstanding reimbursements, which Ben also directly included:
The “in part due to her strongly requesting it” part currently feels unclear to me, and I would probably say something slightly different given my current epistemic state. I genuinely believe that Alice was afraid and concerned that Nonlinear would withhold the €2900 from her, and it is clearly the case that she racked up $3000 worth of reimbursements on a $1000 dollar salary, which is pretty scary. The central point for this inclusion was to demonstrate the financial dependence. Saying instead “After she left she was paid back ~$3000 of outstanding reimbursements, which she was concerned Nonlinear might withhold from her or dispute being legitimate, which did help her financial situation” seems like it gets the same point across and captures my current epistemic state better.
I think it would have been better to say that she was paid back $3000 in outstanding reimbursements instead of salary, but I don’t think it makes a huge difference.
Yeah, I think it was a bad move to not include the follow-up of “some points still require clarification”, but to be clear, I did totally interpret Emerson’s email as saying it was basically a good summary. I think it’s still bad to not have included the next sentence, and I was glad that it was corrected in the comments at the time (and I also somewhat think Ben should have updated the post to clarify that point so the correction is more easily visible).
For context, here is Emerson’s message:
Given the agreement-votes on this comment at the time, other people seemed to agree with me that summarizing the above as just a “Good Summary” was a kind of understandable mistake to make, and that this email did not successfully convey that Emerson disagreed with the points in the summary as provided,
Overall, yeah, given the messiness of a situation like this, I still feel quite good about basically all the quotes that you highlighted. I have now read the whole appendix to this post, spending over 10 hours going through it, and I can’t find any clear rebuttal to almost any of what I consider the core claims in Ben’s post.
Again, I do think someone might point out some inferences together with evidence in the appendix that might make me think some of the information that we quoted from sources is inaccurate. Both memory is fallible, and I also totally think that both Chloe and Alice had strong feelings here that might have clouded their judgement and memory. But as it stands, having spent a lot of time, I overall genuinely expected Nonlinear to be able to present more counter-evidence than Nonlinear was actually capable of providing, and currently think that any aggregate statements Ben made about probabilities he assigns to various behaviors, and the epistemic statuses he attached to statements made by Alice and Chloe, to still be quite well-calibrated and to capture the situation quite well.
It has still been less than a week, and the appendix is really massive, so if someone has clear rebuttals of things Ben said in the post, especially things that are not quotes or paraphrasings of Alice, but things that Ben directly claimed were true or where he expressed his epistemic state, I would be very interested in that. I have found a bit, but overall not very much.
You say: “This is inaccurate. I don’t think there is any evidence that Ben had access to that doesn’t seem well-summarized by the two sections above. We had a direct report from Alice, which is accurately summarized in the first quote above, and an attempted rebuttal from Kat, which is accurately summarized in the second quote above. We did not have any screenshots or additional evidence that didn’t make it into the post.”
Actually, you are mistaken, Ben did have screenshots. I think you just didn’t know that he had them. I can send you proof that he had them via DM if you like.
Regarding this: “As Kat has documented herself, she asked Alice to bring Schedule 2 drugs across borders without prescription (whether you need a prescription in the country you buy it is irrelevant, what matters is whether you have one in the country you arrive in), something that can have quite substantial legal consequences (I almost certainly would feel pretty uncomfortable asking my employee to bring prescription medications across borders without appropriate prescription).”
It sounds like you’re saying this paragraph by Ben:
“Before she went on vacation, Kat requested that Alice bring a variety of illegal drugs across the border for her (some recreational, some for productivity). Alice argued that this would be dangerous for her personally, but Emerson and Kat reportedly argued that it is not dangerous at all and was “absolutely risk-free”
is an accurate characterization of this sentiment: she was asked to pick up ADHD medicine in a place where it was believed not to require a prescription, and bring it back to a place where it does require a prescription, but then was told not to worry about it when it was found that it does require a prescription where she was going to pick it up.
To me, the former does a really bad job of capturing the latter, sounding WAY worse ethically. But I’d be curious to know if others agree with me or if they think that Ben’s paragraph captured this in a fair way.
On most of the other points I mentioned, it seems you feel that Ben made mistakes or ommisions, which I agree with:
“I currently feel a bit sad about the “daily” here, but would write roughly the same thing replacing it with “multiple times a week”.”
“I do think it’s currently sad that Ben’s post didn’t also mention Alice’s company”
“Yeah, I think it was a bad move to not include the follow-up of “some points still require clarification”″
On the taxi situation, I can’t speak to what it is like to get taxis in all the areas they lived, though this is one of those things Ben could have checked rather easily by asking where they were that she couldn’t get cabs and trying to book a cab (I don’t believe he bothered to check, but correct me if I’m wrong). When I stayed with all of the involved parties for a few days: Alice, Chloe, Kat, Emerson, and Drew (which was not in Peurto Rico), I got taxis twice, and it was slightly annoying—I spent about 5-10 minutes explaining where exactly to meet me, but was able to successfully get taxis on both of those occasions. Of course, that was just in one place that they lived (I think they were in that location for a couple of weeks if I recall correctly), and it might have been harder in other places, but at least in that location, getting taxis was no more than a minor nuisance.
Update: since it’s so easy to verify the claim about taxis, I just went ahead and checked it myself. My understanding is that Chloe was talking about Cabo Rojo, Puerto Rico. I used google maps to find and then call three taxi companies in that region. The first two didn’t pick up. The third said to text them, which I did, and they gave me a quote for getting a taxi today to drive 40 minutes (which was $45). It took about 10 minutes of my time (and 20 minutes on the clock, since they took ~10 minutes to give me a quote).
I didn’t realize this earlier, but in their evidence doc, Nonlinear talks about a similar check they did: “Lastly, it was not complicated to get a taxi there. I quickly checked, because sometimes we are in places that are truly remote. But there were three taxi services in the area that could have picked her up or driven her there. I called one of them and they said it would cost $30 and they could come pick me up whenever. And one of the other places where we were living at the time, she literally had to book a taxi for me out there, so I know she knew how and that it was possible.
There just wasn’t Uber there, so she’d have to make a phone call to a taxi.”
I can’t tell if you think Alice gave Ben basically accurate information and didn’t leave out critically important details, or if you think she did leave out critically important details (or directly lied to Ben), but it didn’t matter because Ben’s post was justified regardless.
Sure! DMd you. I might also ping Ben, though want to mostly give him space and time to write a reply and not have to worry about stuff in the comments for now.
No, that’s not what I am saying. I am saying that that paragraph, combined with Kat’s paragraph is a pretty reasonable summary of the evidence, and I think overall I think was pretty good at capturing the gist of the story (like, the two conflicting stories capture my present epistemic state still pretty well, which hasn’t changed very much).
My current model is that Kat is understating the degree to which she was asking Alice to carry drugs across borders, and so your summary (which is basically just a summary of things Kat has said) is also not an accurate summary of my epistemic state.
My current epistemic state is something like “Kat did ask Alice to bring over substantial number of prescription drugs, and probably also some recreational drugs that were illegal to bring across the border. My guess is most of the requests of this type were made in voice. I think the pressure here was substantial enough that I would feel very uncomfortable doing this to an employee of mine, and where if I saw the whole interaction I would think it’s a quite substantial flag, but also that no one did anything so egregious that this thing alone should cause any major repercussions for anyone involved”.
Agree that my epistemic state on this point is also something close to this.
Summarized would be “something like asking her to bring the drugs probably happened, and if so was a mistake that I’d hope was learned from, but the major issue would be if she was pressured to do it, and I’m unsure if I trust the person reporting enough to decide either way without evidence.”
[Edit: I know this is probably a frustrating thing for others to read, but seems worth saying anyway… since making the above comment I’ve had private information shared with me that makes me more confident NL didn’t act in an abusive way regarding this particular issue.]
Ok, I pinged Spencer. He sent me screenshots of text messages he sent Ben that he sent ~2 hours before publication of the post (in the middle of the barrage of comms that Nonlinear was firing off at the time, which included the libel threats), and which Kat posted to the comment thread less than 48 hours after the messages were sent to Ben.
I stand by my summary that everything Ben knew at the time of writing the post, made it into the post. Of course if you send something 2 hours before the post is published, late at night, it’s not going to make it into the post (but it might very well make it into a comment, which it did).
Ben made a bunch of other changes the day of publication. I know that because I pointed out errors in his post that day, and he was correcting them based on me pointing them out (e.g., all of his original quotes from glassdoor that he claimed were about Emerson were not actually about Emerson, which he didn’t realize until I pointed it out, and then he rushed to find new quotes to correct it). I’m sure he had a lot on his mind at that time, so I don’t think it’s egregious that he didn’t add mention of the fact that he had screen shot counter evidence about the “no food while sick stuff”, but it clearly seems to me to be a mistake on his part to not adjust the post or at least acknowledge it in the post. And I know he received the screenshots because he acknowledged getting them. You’re saying it made it into a comment as though Ben gets credit for that—but wasn’t it Kat who posted that comment? He also chose to rush the post out that night despite knowing there was counter evidence. I was honestly shocked he was trying to rush the post out that night because of all the errors I was finding in his post, which I expressed to Ben that day.
Update: I only just saw this point you made, including here for context and because it helps answer what I said: “since publishing this post required coordinating with many (5-10) external sources and witnesses, with many of them having a strong preference for a concrete time for the post to be published and they can plan around, so they can get ready for any potential retaliation and fallout. “
Update 2: “Spencer and Nonlinear knew the claims we were planning to put into the post on this matter roughly a week in-advance.” I’m not sure what you’re basing this on, but this is not accurate. I saw the post on Sept 6 for the first time and Ben published it early morning of Sept 7. During that short gap I sought to help him correct some errors in the post, which he did.
Update 3: the post was indeed being edited in meaningful ways on Sept 6, I know because I was helping Ben identify mistakes he had made and he was making changes based on that (such as the glass door misattribution)
Sorry, I had assumed that Nonlinear had shared with you the document we shared with them before publication of the post. My Slack records indicate Ben had two calls with you in the week before publication, and my guess based on the feedback I see from you paraphrased and copied in the Slack, is that you were aware of the claims in the post.
I might be wrong here and the Slack records could just align in a kind of confusing way (I can’t find the date of your first call with Ben, but am confident there was one more than 24 hours before publication), in which case I apologize.
Yes, Ben was making changes the day of the publication, I don’t think I said otherwise?
I also think sending something 2 hours before publication is again different from that (like clearly we can at least agree that if you had sent it 15 minutes before the publication time that it would not have been reasonable to say that Ben had access to information during the writing of the post that didn’t make it into the post?).
I really would not describe the post as being “rushed out”. The post had been worked on for over 1000 hours. I also think you are overstating “all the errors you were pointing out”. You pointed out two things which to me still seem relatively minor.
I think if Kat hadn’t posted the screenshots in a comment, Ben would have left a comment or edited the post. We really tried pretty hard to include anything that was sent to us, and I think Ben managed to include a lot of information and epistemic nuance in the post, while still maintaining the basics of readability and clarity.
Is it just me or does the number keep going up with every retelling?
When we did a postmortem on it, somewhat over 1000 hours is how high the total staff cost seemed to us, and that was a few months ago.
I think it’s totally plausible that in a few places I or someone else on the team used a lower number that they felt more confident in. In-general the structure of “over X” is something I usually use when I am not sure about X, but want to give a quick lower bound that allows me to move ahead with the argument, so it seems totally possible that in another context I would have said “multiple hundreds of hours” or “300+” hours or something like that, because that was enough to prove the point at hand.
Edit: Oh, I see the links now, didn’t see them when I first wrote the comment.
I think the key difference with that quote and my number is that it just includes Ben’s time, as opposed to total staff time. For example, it omits work done by anyone else on the team (which roughly doubles the total amount of time spent, spread across me, Robert and Ruby), as well as others who we’ve brought on board to help with the post (we worked with 2-3 external collaborators who ended up pairing with Ben for multiple weeks).
My guess is also Ben’s number is a bit low for his own time spent on it, though I think we are now getting into definitions of what counts as “working on it”. We don’t have detailed time tracking, so this is a bit hard to operationalize, but my guess is if you added up all the staff time of Lightcone staff and external collaborators, and removed the project of writing the Nonlinear post, you would indeed end up with somewhat more than a thousand hours of additional free time across those people.
Thanks. I didn’t mean my comment to come across as a “gotcha” question fwiw (not saying that you said it was a gotcha question, but I realized after I commented that it’d be a reasonable interpretation of my comment).
For what it’s worth, I find it extremely plausible that a post like this both took an inordinately large amount of time, and that people will systematically underestimate how much it took before they started doing more accurate time-tracking.
It does seem very sad that the voting on this post seems a bit broken (it also seemed broken on the original Nonlinear post). Like, do people think I am lying about the amount of hours it took? I would be happy to provide the data that I have, or have someone else who is more independent to the Lightcone team provide an estimate. It seems very weird to downvote an answer to a straightforward question like that.
Hmm, well Ben said “(for me) a 100-200 hour investigation” in the first post, then said he spent “~320 hours” in the second. Maybe people thought you should’ve addressed that discrepancy?️ Edit: the alternative―some don’t like your broader stance and are clicking disagree on everything. Speaking of which, I wonder if you updated based on Spencer’s points?
Apologies, that was my fault. I wrote the comment and then I realized that I was demonstrating poor reasoning transparency, so then I hunted down the relevant links. My guess of chronology was that I had the hyperlinks added in after you started commenting, but before your reply was visible.
Sorry if that burned extra time on your end. :)
Ah, cool, I was really surprised when I saw the links on refresh, but they fit so naturally into the comment that I thought they clearly must have been there in the first place.
No worries, it cost me like 3 minutes.
This would make sense to me if Ben had been working to an external deadline, but instead this is directly downstream from Ben’s choice to allocate very little time to draft review and ensuring he had his facts right. It sounds like Spencer sent these text messages <24hr after being sent the draft; how quickly would he have needed to turn around his review to count?
To be clear, we were working to a substantial degree to an external deadline, since publishing this post required coordinating with many (5-10) external sources and witnesses, with many of them having a strong preference for a concrete time for the post to be published and they can plan around, so they can get ready for any potential retaliation and fallout.
There was wiggle room in that date and time, but by the time Spencer sent this, the post and publish-date was really quite locked in.
I think 24 hours before publication would have been enough to include them. Maybe even 12 hours. As I mentioned in other places, we did send Nonlinear (and Spencer) a list of the relevant claims in the final post, including this one, so I think the fact that the literal draft was only shared 24 hours in advance is irrelevant. Spencer and Nonlinear knew the claims we were planning to put into the post on this matter roughly a week in-advance.
For example, in the call that Nonlinear cancelled with us a day before publication, that would have been a pretty good time to share such evidence with us, and if they had given additional evidence then, it would have made it into the post.
But separately from that, I am not sure what you mean by “count”. Spencer claimed that “we had screenshots that didn’t make it into the post”. I think a reasonable reader would infer from that when the post was being written, we had access to those screenshots. By the time Spencer sent these screenshots, the post was no longer being written in any meaningful way.
I don’t really see how this is a defense. The fact that you have promised some third parties to do X does not justify you in doing X if X would otherwise not be morally acceptable. And publishing harmful statements about someone that you have good reason to think are false does not seem morally acceptable.
Yes, this does seem like deciding in advance what side you’re on and who deserves consideration like determining when the post goes up.
It is a defense that in as much as I think anyone working on a post similar to this, mostly independently of skill level, would end up having to make promises to sources of this type, in order to be able to share concerning information publicly.
Of course, if you think posts of this whole reference class are bad, and it was bad for us to even attempt to make a post that tries to publicize the extensive rumors and concerns that we heard about Nonlinear, then I think it’s not a defense.
But if you think people should attempt to spread that kind of information and share it with more parties, then I think this will somewhat inevitable come with constraints like having to keep publication deadlines and coordinating the many stakeholders involved in such a thing.
Like, what is the alternative that you propose we should have done instead? Not made any promises to our sources at all about doing things that protect them from retaliation and limiting the costs on them? I think in that case you don’t get to talk to sources, or you only get to do it for a bit as people get burned and hurt and stop talking to you.
I am pretty sure Ben has published no harmful statements about someone that he thought were false. Indeed, as I have said many times, he seems to have been exceptionally careful with the epistemic states he attached to his statements in his post.
I’m well aware of the difficulties of balancing competing stakeholders giving you feedback late on posts and trying to hit publication timing targets. I think you had several valid options:
Never make commitments about publication date and time in the first place.
Make commitments, but be clear they are provisional. When you receive this information, email your sources saying “hey guys, really sorry but we just received some last-minute info that we need to update on. We’ll circle back to coordinate a new launch date that works for you.”
Give Spencer a reasonable deadline to respond, committing to take into account feedback received before this deadline.
Delete that section and publish on the original schedule.
Edit the section and publish on the original schedule.
I mean, to be clear, we did this the first time Nonlinear disputed the relevant section.
I think this section is really quite clear. We have one report from Alice saying that she quit being vegan. We directly include, in the next paragraph, the fact that Nonlinear disputes this. I really don’t think we misled anyone. The screenshots sent did not actually materially change anything in the paragraphs above, indeed both of the paragraphs are still fully accurate (and in as much as Alice claimed that she did not get food while indeed getting food, that is IMO an important part of the story that seems important for other people to be able to cross-check).
I think the choice of “you have some sources, you cite the sources while being really quite clear that you don’t fully trust your sources, and when a thing gets directly disputed by another source you say that directly” is a reasonable thing to do. Again, as I’ve said an enormous number of times, we never had an intention of fully litigating all of these claims before publication, which would have been completely infeasible time-wise.
The alternative to Ben’s post would have probably been a series of fully anonymous posts with extremely vague high-level accusations that would have been extremely hard to respond to. We tried to make the claims concrete and provide an interface to aggregate information at all.
Like, what kind of edit would you have preferred us to do instead of the above?
I strongly disagree. Alice’s and Nonlinear’s perspectives are portrayed with very different implicit levels of confidence in those paragraphs. Alice’s perspective is stated as a fact—“nobody in the house was willing to go out and get her vegan food,” not “Alice says nobody in the house was willing to go out and get her vegan food.” In contrast, Nonlinear’s perspective is shared as “[Nonlinear] says [x].”
I think most readers who trust Ben to be truthful would assume, from the way those paragraphs were worded, that Alice had much better evidence to support her claims, and that Nonlinear was doing some slightly deceitful reputational management by countering them. But that isn’t what turned out to be the case:
Nonlinear has evidence that on December 15, they had oatmeal, peanuts, almonds, prunes, tomatoes, cereal, an orange, mixed nuts, and quinoa (which Kat offered to cook) in the house.
On the same day, Kat had successfully purchased mashed potatoes for Alice.
On the next day, they apparently went out and purchased both Panda Express vegan noodles and vegan burgers for Alice.
At some point, Emerson went out and tried to purchase Alice more food despite his knee injury, but he couldn’t find the very specific items she requested.
Then, on December 18, it looks like Alice’s first non-vegan meal was a vegetarian pizza she ordered (rather than non-vegan food already in the house). It looks like she ordered it right after Kat reminded her that they already had vegan noodles in the fridge.
On top of all of this, apparently everybody in the house was either sick or injured, but Ben’s post only mentions that Alice was sick.
It seems that Alice/Ben have no evidence to counter any of the points above.
So the original claim that was stated as fact (“nobody in the house was willing to go out and get her vegan food”) seems very wrong. Which is sad, because it’s a very serious accusation that most people would assume was not made lightly.
Yeah, sorry, I think I was too strong in my language above, though my sense is you are also interpreting my answer to be about a somewhat different question than the one I perceived Larks as asking. To clarify where I think we are on the same page: I am pretty unhappy about that section, and wouldn’t ask Ben to write something different given what I believe today.
The thing I was responding to was whether we misrepresented the evidence that we did have at the time.
On that topic, I do think it was a mistake to omit as many of the “Alice/Chloe claims that X” in the post as we did, and fall back into a neutral third-party way of summarizing the claims, and given that we did, I think it makes sense to hold Ben and Lightcone more responsible for the veracity of statements that did not include an explicit “Alice/Chloe alleges X”.
I also think that there is a pretty reasonable case to be made that we should have waited longer on getting more evidence from Nonlinear. I felt conflicted on this topic then, and feel conflicted now. I really hate that the situation we were in made it quite hard for us to wait longer for Nonlinear to respond to us. I am still not fully sure whether I would wait if I was in this situation again, since the considerations against waiting were also quite strong, though overall I am leaning slightly that waiting would have been the better option (I do not think this forgives or excuses Nonlinear’s attempts at intimidation and threats of retaliation).
However, overall on the question of “did we accurately summarize the evidence available to us”, I think Ben’s post and this section is doing pretty well.
I agree that we frame Alice and Chloe’s evidence as more trustworthy, and in-aggregate, across the whole post, I stand behind that framing, in that I think Alice and Chloe are substantially more reliable sources of evidence than Kat and Emerson. I agree that in this situation I think this went the wrong way around and it looks to me like the vegan food situation seems like it was represented to us in a substantially misleading way, and I am still hoping for me or Ben to follow up with Alice on this topic and figure out whether I am missing something. However, I think on-average the framing of the post was not misleading about the balance of evidence that we had received to that point (including accounting for expected future evidence Nonlinear that we expected Nonlinear might provide).
Some smaller nitpicks on your comment:
It’s true that we don’t share Nonlinear’s perspective with the same authority as Alice and Chloe’s. We did also include a summary directly written from their perspective, which I do think helps:
Another quick comment:
I would give people some time before concluding that. While Ben (and I) are trying really hard to not be dragged into a full-on follow-up investigation of this, I do expect there will be some kind of response to this which includes procuring more evidence. I personally do feel pretty convinced on this point, but I am not updating on Alice or Ben not providing more evidence in coming to that conclusion, since they haven’t responded to anything so far, and I do know that many of the claims in the OP and associated appendix are inaccurate, and those also haven’t been responded to yet (it includes many claims about what Ben believes or what the process of writing Ben’s original post was like, which I am very confident are inaccurate).
Given Chloe is not involved in this claim, do you also stand behind the framing that Alice is more reliable than Kat/Emerson?
I am substantially less confident in that claim, though yeah, I would still overall say I believe it (it’s not super well-operationalized so not super clear what a probability would mean, but like, I guess I am at ~80% that if I knew all the facts and had arbitrary insight into Alice’s, Kat’s and Emersons’ life that I would overall expect Alice to be reporting more accurately than Kat and Emerson)
I’m not sure if Spencer sent you all the screenshots or just some of them, but something along the lines of:
The screenshot Ben received at the time is one of the ones that Kat linked in this comment:
https://forum.effectivealtruism.org/posts/32LMQsjEMm6NK2GTH/sharing-information-about-nonlinear?commentId=Ejbe8ukX6FhrfRv5C
Importantly the screenshot only covered events on December 15th.
Here is the relevant screenshot:
Kelsey Piper in the thread summarizes these screenshots (together with some other screenshots that Kat shared) as follows:
My guess is this aligned with Ben’s interpretation at the time. The screenshots were relevant evidence, but they did not directly disprove anything in the original article.
Kat then shared further screenshots in the comments, which importantly were not shared with Ben beforehand (unless Spencer failed to forward them to me in my DM with him yesterday), that demonstrated that on the next day Kat did successfully bring her food.
However, the story, in the above screenshot, on December 15th, is that indeed Alice did not get food, despite her requesting it. The screenshots that Spencer sent us appear to fail to include the most relevant part of the conversation, which is that they did indeed fail to get her vegan food
that dayduring that trip.(Edit: Kat disputes this below, sharing some additional screenshots that seem to show that Kat did get food for Alice later that day, which seems important to get right. Though I don’t think Spencer’s screenshot demonstrated this).
Here are the edits I currently agree would have been better, though I think they are minor enough that I don’t currently see it as a major error to not have included them:
I really encourage you to look at the screenshots, Kelsey’s summary, and Kat’s original comment on the Nonlinear post and explain to me how these screenshots falsify part of the post. As we later received more screenshots, it seems like we actually received confirmation that the conversation on that date did indeed not result in Alice getting food.
(Edit: Kat shares some additional screenshots below that do seem to show Alice got food on the 15th, though not from the restaurant trip that was talked about in the screenshot Spencer sent us)
I’m a little bit confused about Kelsey’s summary—it contains a line about rejecting burgers because they were ‘fast food’ that doesn’t seem to be in the original. So I don’t think it can reflect Ben’s state of mind in that way.
If you only had the one screenshot (9:53 to 10:28 timestamps), I agree that you can’t infer that Kat cooked for ‘Alice’, nor is there proof that the discussed burger trip actually took place, though I think they strongly imply it will—certainly Alice seems to think it has been agreed and will occur. However, I find your comment about 15th vs 16th unconvincing because ‘Alice’ explicitly claims a 2 day duration, so food the next day would also contradict this (assuming the 15th is the first day).
Here is another possible version that reflects just the one screenshot:
I think my key objections to the original version is asserts as fact that Nonlinear refused to get her food, vs this being an unverified claim, and that it does not reflect that NonLinear didn’t merely dispute it, they offered evidence.
Just to be clear, that burger trip did indeed not happen that day, if I understand it correctly. What instead happened is that Kat went out a few hours later and got Alice mashed potatoes at a store (which is not really hinted at at all in the screenshots).
Yeah, I think this version is reasonable and I would have preferred to post this version (and somewhat think that we should have updated it ASAP, even after publication).
This is false. Alice got food on December 15th. She got food 2.5 hours after she asked. Actually, she never asked me, I just offered when it seemed like she was struggling.
It says December 16 at 12:14am because I was in Europe at the time, so it’s showing the European time zone. It was Dec 15 at 7:13pm in the local time when this occurred.
She brought up being hungry at 4:53pm. I immediately offered to cook her the food in the house. When she didn’t want any of the food in the house or food from any non-fast food restaurant within a 12 minute drive of home, I went out, while sick myself, and got and cooked her food. The only vegan food that fit her criteria in the store.
The only complaint she can legitimately say is that we did not get her Panda Express as fast as she would have liked (we got it for her the next day). She waited 2.5 hours for food. And she could have had it sooner if she’d wanted any of the food in the house, which she usually ate nearly daily and enjoyed. She just didn’t want that food. She wanted fast food and didn’t get it as fast as she preferred.
I’m currently back on the same time zone, so here’s the same screenshot, but showing the right time zone dates and times
Thank you! This definitely seems like highly relevant evidence.
Can you clarify whether Kelsey’s summary of the December 15th conversation is accurate or inaccurate? It’s totally possible that I am misreading the screenshots, though my best interpretation was indeed the interpretation that Kelsey made in the screenshots.
I would be happy to correct the statement above if I am wrong here.
I do think this issue seems somewhat separate from the question of “did the screenshots that were shared with us materially affect the things Ben wrote?”.
To be clear, this is relevant in as much as the original screenshot was evidence of there being more things you could share here, though I currently maintain that I don’t think the screenshots that were shared with us showed any material error (given that Kelsey also walked away with the same impression of them being consistent).
I also totally care about just setting the record straight and getting the object-level issue right here, and in as much as there isn’t anything very weird going on with the screenshots you sent, I think you provided pretty decent proof here and am changing my mind on the December 15th issue (and think if you had shared those screenshots with us instead, I think it’s pretty likely Ben would have somehow made sure that they made it into the post).
Kelsey’s summary was wrong in a number of important ways.
She missed the fact that we did indeed succeed in getting her vegan food (I found at the nearby store, despite being sick myself). 2.5 hours after we first offered. And it would have been faster if she’d wanted any of the food in the house, or chosen a restaurant that had vegan options for Emerson and Drew to go to.
It doesn’t mention the vegan food that was in the house already that I offered to cook (Alice ate oatmeal almost every day and she loved quinoa. Later when I cooked some up for her, she loved it, like usual, cause quinoa is the Queen of All Foods).
It doesn’t mention that Drew said he would go to any restaurant within a 12 minute drive from our place and she just… didn’t choose a restaurant. She only wanted fast food. So they ended up choosing a restaurant that didn’t happen to have vegan options aside from the usual fries.
A quick look at Google Maps shows that there was over 20 restaurants that fit that criteria in the area. It wasn’t restrictive at all.
She frames it as they didn’t get her the food she wanted “because they [didn’t] want to get fast food.” It’s important to note that Emerson and Drew also are people whose preferences matter. Just because Alice is sick doesn’t mean everybody has to drop their own needs and preferences to get her the very particular food she wants.
She frames it as Emerson and Drew being somehow inconsiderate and shallow, when you could just as easily frame it as Alice not considering the needs or preferences of anybody but herself, expecting everybody to drop everything and go out of their way so she can get the very specific fast food she wants. Then, when she doesn’t get exactly what she wants as fast as she wants, she goes around telling lies about what happened to destroy a charity (e.g. nobody willing to get her food)
As for what Ben knew before publishing, if you look at that screenshot, you can see that:
Drew has offered to pick her up food (“Drew suggested he could otherwise pick up stuff”)
Me and Emerson offered to pick her up food (“Me and Emerson can do it if he can’t”)
I offered to cook her the food in the house (“Could make you some quinoa”)
Ben said in his post that “nobody in the house was willing to go out and get her vegan food”. This is absolutely false. Ben had seen this screenshot clearly showing that we were willing to go out and get her food.
Note that she didn’t even ask for food. I just offered because I could see she was in need (“want me to order food?”)
It also said “so she barely ate for 2 days”. It shows in the messages that there was vegan food in the house. So that was clearly not the reason she didn’t eat for days.
She had plenty of options but she wanted fast food in particular.
We got her her first choice of fast food the very next day (remember, she only started asking for food in the evening the previous day. It was also hard to get stuff for her. This was our first experience with covid and we were trying to figure out how to manage it, try to have it not spread, etc etc. It was quite a stressful and overwhelming time.).
Because she didn’t get a very particular fast food as fast as she wanted, she interpreted this as us being heartless people who wouldn’t take care of a sick person in need. She told Ben a false and misleading story about us not being willing to go out and get food for her.
I’m waiting for Ben, or someone else, to make a table of claims, counter claims, and what the evidence shows. Because nonlinear providing evidence that doesn’t support their claims seems to be a common occurance.
Just to give a new example,Kat screenshots herselfreplying “mediating! Appreciate people not talking to loud on the way back[...] ” here, to provide evidence supporting that there was not a substantial discussion that occurred. However, I can only interpret the use of “mediating!” to indicate that there was in-fact a substantial amount of discussion at play.Edit: Retracted as correctly pointed out by @Sean_o_h , I read meditation as mediation.
Uh, the word in that screenshot is “meditating”. She was asking people to not talk too loudly while she was meditating.
That is correct.
Oh thanks for flagging, I will retract it now
Can you clarify what you mean by “very little time”? Haybrka reports spending 1000+ staff hours, and even Ben’s much more conservative estimate of 100-200 hours doesn’t feel fair to me to describe as “very little”
Sorry, I’m trying to talk about the amount of time for ‘adversarial’ fact checking: when Nonlinear knew the accusations and could provide specific counter evidence. I agree he put a ton of time into the project overall.
Just a note that standard practice on these kinds of jobs is that you get a credit card to make purchases with, and are never using your own money that is later reimbursed.
A big reason for this is the massive mismatch in what money is worth. Employers might think covering a $100 grocery trip until you get reimbursed is not a big deal, but to an employee that might have been their own food money or rent.
The standard answer is you either let your employee borrow your credit card, or you give them their own credit card. You can put a lower limit on it to protect yourself, and can also see the credit card statement (which can be paired with receipts if you don’t trust them not to add on extras. I was always careful that my families get all the receipts but they generally just threw them away because they trusted me)
My model has been there should be social enforcement for both poor epistemic practices and rude/unkind communication.
I have been an active commenter in both posts, with a goal of social pressure in mind (i.e. providing accountability and a social pressure to not behave inappropriately towards/with your employees).
I’d be interested to hear meta level criticisms of my approach (e.g. “social pressure is inherently bad”). Because, whilst I don’t want witch hunting that employs poor epistemic practices, I do think social pressure plays an important role in stabilising communities. Perhaps someone can change my mind on this? If you do change my mind, I’ll certainly comment a lot less.
To me it seems like everyone individually applying social pressure is hard to calibrate. Oli seems to be saying that he and Ben did not intend the level of social consequences NL has felt based on what they shared, but rather an update that NL shoudn’t be a trusted EA org. I think that it’s hard to control the impression that people will get when you provide a lot of evidence even if it’s all relatively minor, and almost impossible to control snowballing dynamics in comment sections and on social media when people fear being judged for the wrong reaction, so it just might not be possible for a post like Ben’s to received in a calibrated way.
This sounds right, but the counterfactual (no social accountability) seems worse to me, so I am operating on the assumption it’s a necessary evil.
I live high trust country, which has very little of this social accountability, i.e. if someone does something potentially rude or unacceptable in public, they are given the benefit of the doubt. However, I expect this works because others are employed, full time, to hold people accountable. I.e. police officers, ticket inspectors, traffic wardens. I don’t think we have this in the wider Effective Altruism community right now.