Practically A Book Review: Appendix to "Nonlinear's Evidence: Debunking False and Misleading Claims" (ThingOfThings)

post by tailcalled · 2024-01-03T17:07:13.990Z · LW · GW · 25 comments

This is a link post for https://thingofthings.substack.com/p/practically-a-book-review-appendix

Subtitle: Taking nonprescription amphetamines across the U.S.-Mexico border is a felony

I haven't followed the controversy long enough to be able to tell how correct Ozy is, but what I like about this post is that I find it better structured than some posts, and with better focus on strategically relevant information.

For instance, in response to Nonlinear's claim that they weren't isolating Alice and Chloe, Ozy wrote:

Kat believes in the importance of digital nomads remaining socially connected to others. However, Kat and Emerson had a consistent pattern of encouraging Alice and Chloe to spend time with people they considered high value (i.e. effective altruists, especially those working in AI safety) instead of people they considered low-value. To be clear, Kat and Emerson didn't think Alice and Chloe should completely isolate themselves from people who weren’t effective altruists. Kat encouraged Alice and Chloe to call their families regularly. She explicitly supports spending some time with locals. Friends and family who didn't work in AI safety were invited to travel with Nonlinear, although they were lower priority to invite than AI safety people.

However, the vast majority of Kat’s evidence that she didn't isolate Alice and Chloe is evidence that she didn't isolate Alice and Chloe from effective altruists, particularly "top" effective altruists working in AI safety. Alice and Chloe were given lots of access to so-called top effective altruists: there was an average of 7 people living in the house. Nonlinear encouraged networking with FTX people. They traveled with Chloe's boyfriend, whom Kat Woods considered to "have high potential." Inviting people to travel with Nonlinear was framed as "one of the highest ROI things you can do”—that is, as an important means of bettering the world.

Kat and Emerson discouraged Alice from visiting her family because her trip overlapped with "some of the top figures in the field" coming to visit. (The chatlogs are suggestive that Alice timed her visit around a family emergency, but Kat doesn't explicitly mention this.) Kat also discouraged Alice from spending too much time socializing with locals, saying that she would have higher impact if she spent time with higher-value people.

It doesn't seem to have occurred to Nonlinear that Alice and Chloe might have been complaining about being isolated from people who aren't effective altruists.

I was thinking along similar lines when I was reading this part of Nonlinear's response, but I hadn't followed it well enough to make the argument in detail.

When I say "strategic", what I mean is that the above information focuses on the specifics of the relationships and the roles that people had. This is as opposed to "generic" meanings of "social isolation", which might simply be defined based on the presence or absence of other people.

(As a side-note, I think there might be a general rationality phenomenon to study here? The distinction between strategic phenomena and generic phenomena, and miscommunication/deception when one is substituted for the other. But this gets a bit off-topic for the overall post.)

25 comments

Comments sorted by top scores.

comment by Jiro · 2024-01-05T17:04:24.312Z · LW(p) · GW(p)

If the accusation is "they isolated me", showing that she wasn't isolated rebuts it. If the accusation was "they isolated me, but only from some people", she should have said that in the first place.

If you let someone make a wide scope accusation, and then the accusation is proven false, and then you say "well, they didn't rebut this narrower scope accusation", you are enabling dishonesty. (Especially if the narrower scope accusation doesn't sound as impressive as the original one.) People can only rebut accusations that were made and it's easy to just keep changing the accusation to escape being proven wrong.

Replies from: tailcalled
comment by tailcalled · 2024-01-05T17:19:16.669Z · LW(p) · GW(p)

it's easy to just keep changing the accusation to escape being proven wrong

This doesn't seem true. It should be clear why strategic isolation (i.e. isolation from people who aren't specifically preferred by Nonlinear) is a problem. Meanwhile, if strategic isolation was disproven and some other forms of isolation were raised (e.g. isolation for medical reasons to avoid contagion), these other forms of isolation would either be justified or obviously not exist, and therefore wouldn't be possible to bring up.

Strategic meanings of words are uniquely relevant in the context of abuse, because most of the damage caused by abuse and most of the means used to abuse are in the strategic domain. It is therefore especially legitimate to be talking strategically rather than generically in the case of abuse.

Replies from: Jiro
comment by Jiro · 2024-01-05T19:08:29.661Z · LW(p) · GW(p)

Strategic meanings of words are uniquely relevant in the context of abuse

"You can only rebut the accusation that someone actually makes" is relevant in all contexts.

The standard that you seem to be suggesting is Kafkaesque. Someone accuses you of something, you prove them false, but that doesn't count because of strategic meanings of words. What?

(That's aside from the question of how strategic meanings of words are even relevant. Nonlinear had no control over the words in Ben's post.)

these other forms of isolation would either be justified or obviously not exist, and therefore wouldn’t be possible to bring up.

The problem is changing the accusation after the fact, not the details of the change. There are plenty of other ways the accusation could be changed after the fact that aren't about forms of isolation.

Replies from: TekhneMakre, tailcalled
comment by TekhneMakre · 2024-01-06T08:53:38.901Z · LW(p) · GW(p)

The standard that you seem to be suggesting is Kafkaesque. Someone accuses you of something, you prove them false, but that doesn't count because of strategic meanings of words. What?

But imagine this from the other side of a conflict. There's a social norm:

Don't isolate people (e.g. because it makes them vulnerable, e.g. to abuse).

Now a hypothetical (cartoonishly explicit) bad actor comes along and says "Aha, I know what to do, I will use my soft power to isolate my employee, but only from some people, and that way I'm not "isolating" them, but I can still control their social context of influences, support, and ideology". (To be extra clear: I'm not following the story in detail and I'm genuinely not claiming that Nonlinear is like this; there's some possible relevance, in that their genuinely well-intended actions might possibly have had a similarly bad effect as this hypothetical cartoonish bad actor would hypothetically have had.)

So this bad actor does this. Now, did they isolate the person? Did they violate the norm? Can you accuse them of isolating their employee? Do you have to exactly specify what shape of isolation, on pain of making an infinitely malleable accusation? If you later specify the shape / form of the isolation, are you changing the accusation?

Replies from: Jiro, tailcalled
comment by Jiro · 2024-01-06T23:23:22.575Z · LW(p) · GW(p)

You should be specifying enough so that they don't say something that rebuts the accusation and you can then respond with "well, they rebutted what I actually said, but they didn't rebut what I meant, so it doesn't count".

Alice and Chloe report that they were advised not to spend time with ‘low value people’, including their families, romantic partners, and anyone local to where they were staying, with the exception of guests/visitors that Nonlinear invited. Alice and Chloe report this made them very socially dependent on Kat/Emerson/Drew and otherwise very isolated.

If that's what she's accusing people of, you have no business later saying "well, actually, they invited her romantic partner, and she was encouraged to invite friends and family, but the accusations are still totally true because she was socially isolated." That's not just a slightly different interpretation of her words, that's flat out saying that the very things brought up in the original accusation as a smoking gun suddenly don't matter now that they were proven false.

comment by tailcalled · 2024-01-06T13:09:50.910Z · LW(p) · GW(p)

I think this is a good first step towards understanding it. That said, this almost frames it as a way to handle adversarially chosen edge-cases, which I think doesn't get to the core of the understanding. One thing I would highlight:

Don't isolate people (e.g. because it makes them vulnerable, e.g. to abuse).

That is, "don't isolate people" isn't a rule because People Are Happier When They Are Around Others (even though that is true in a generic sense, and happiness does in a generic sense correlate with goodness). Rather, "don't isolate people" is a rule because of the strategic consequences of isolation. As such, it is natural that "don't isolate people" would focus on the strategic facet of isolation.

comment by tailcalled · 2024-01-05T23:21:51.766Z · LW(p) · GW(p)

Your position makes sense if you are highly interested in the generic meaning of the word, and not at all interested in the strategic meaning of the word. After all, language is what we make it, and if people decide to make language in such a way that "isolating" is never used in the strategic sense, then of course it would be insanely confusing for someone to insist on using it in that sense, especially in a public accusation of all places. However I think many people, probably even most people, would be interested in the strategic meaning of the word, and not the generic meaning of the word.

Replies from: Jiro
comment by Jiro · 2024-01-05T23:53:16.200Z · LW(p) · GW(p)

After all, language is what we make it

No, it isn't. If you make an accusation about and to third parties, it's those third parties' understanding of language that is relevant--not what "we" understand.

"I have a different meaning of those words I put in the accusation because language is what we make it" is dishonest.

Replies from: tailcalled
comment by tailcalled · 2024-01-05T23:59:48.401Z · LW(p) · GW(p)

Sure, but since both my and Ozy's reaction to the counterarguments about isolation was "it sure seems like there could still have been some isolation", there are at least some third-parties who understood the strategic sense. I'm not making the argument that Alice and Ben can use the word however they want, I'm making the argument that lots of people understand this.

Maybe we could test it. Like we could give some people a story of some cult which insists that some victims spend all their time interacting with cult members, and then ask the people whether the cult isolated those victims (as would be implied if people use the strategic meaning of the word) or did the opposite of isolating them (as would be implied if people used the generic meaning of the word).

Replies from: bec-hawk
comment by Rebecca (bec-hawk) · 2024-01-06T08:04:12.378Z · LW(p) · GW(p)

I think calling this a strategic meaning is not that helpful. I would say the issue is that “isolated” is underspecified. It’s not like there was a fully fleshed out account that was then backtracked on, it’s more like: what was the isolation? were they isolated from literally everyone who wasn’t Kat, Emerson or Drew, or were they isolated /pushed to isolate more than is healthy from people they didn’t need to have their ‘career face’ on for? We now know the latter was meant, but either was plausible.

Replies from: tailcalled, Jiro
comment by tailcalled · 2024-01-06T08:13:38.528Z · LW(p) · GW(p)

It's unhelpful when it comes to just explaining the meaning of "isolation" as that is better communicated by something like Ozy's description, or in various other ways.

However, I think at least in theory it can be helpful for understanding where to focus one's attention in the case of community drama like this, including for other questions than just the "isolation" one. Maybe not in practice because I haven't gone in depth for explaining what I mean by "strategic" though.

comment by Jiro · 2024-01-08T16:47:18.973Z · LW(p) · GW(p)

I gave a quote from Ben's post above.

She specified families, romantic partners, and locals.

Maybe this doesn't count as fully fleshed, but the new interpretation certainly backtracks on these.

Replies from: tailcalled
comment by tailcalled · 2024-01-08T17:31:34.090Z · LW(p) · GW(p)

How does it backtrack?

Replies from: Jiro
comment by Jiro · 2024-01-09T00:14:46.984Z · LW(p) · GW(p)

Because she specifically said she was isolated from them. She wasn't. And now we're supposed to believe that she really meant a form of isolation where those don't count, even though she explicitly said they count.

comment by Elizabeth (pktechgirl) · 2024-01-03T20:58:26.445Z · LW(p) · GW(p)

Can I suggest cross-posting this to EAF? Looks like it's not there yet. 

Replies from: tailcalled
comment by tailcalled · 2024-01-03T21:11:29.152Z · LW(p) · GW(p)

It would make sense to do so, but I'm neither the author of the post nor very active on EAF, so I'd encourage someone else to do it.

Replies from: sil-ver, pktechgirl
comment by Rafael Harth (sil-ver) · 2024-01-03T23:19:13.092Z · LW(p) · GW(p)

I'm also not very active, but I'm not mentioned in the post and it's been over 2 hours, so I just went ahead and posted it [EA · GW].

comment by Elizabeth (pktechgirl) · 2024-01-03T22:18:05.698Z · LW(p) · GW(p)

I'm thanked in the post so I'm gonna give it a little more time for someone else to post. 

Replies from: bec-hawk
comment by Rebecca (bec-hawk) · 2024-01-06T08:04:40.954Z · LW(p) · GW(p)

Why does that mean you shouldn’t post it?

Replies from: localdeity
comment by localdeity · 2024-01-06T08:37:47.808Z · LW(p) · GW(p)

I'd guess modesty / non-self-promotion norms, and/or avoiding the appearance of a conflict of interest.  "Hey, everyone, listen to this important and wise thing Bob said!  (BTW Bob says that I'm great, and he is so right.)"

Replies from: pktechgirl
comment by Elizabeth (pktechgirl) · 2024-01-23T21:58:06.024Z · LW(p) · GW(p)

mostly I didn't want to be double counted when people assessed support for the post

comment by Algon · 2024-01-03T21:29:07.616Z · LW(p) · GW(p)

When I say "strategic", what I mean is that the above information focuses on the specifics of the relationships and the roles that people had. This is as opposed to "generic" meanings of "social isolation", which might simply be defined based on the presence or absence of other people.

(As a side-note, I think there might be a general rationality phenomenon to study here? The distinction between strategic phenomena and generic phenomena, and miscommunication/deception when one is substituted for the other. But this gets a bit off-topic for the overall post.)

I'm confused. Is specific vs generic the same dichotomy as concrete vs abstract? 

Replies from: tailcalled
comment by tailcalled · 2024-01-03T21:49:13.950Z · LW(p) · GW(p)

The generic meaning is abstract, but the strategic meaning is one of probably several different ways in which things can be concrete. I did not make a specific vs generic dichotomy.

Replies from: Algon
comment by Algon · 2024-01-03T21:55:46.005Z · LW(p) · GW(p)

It sounds like you have some sub-concepts you split "concrete" into w/o being domain specific, but I'm not sure what those sub-concepts would be. If you've written this up anywhere, I'd appreciate a link. 

I'm not actually sorry for going off-topic. This sounds way more interesting than some EA drama.

Replies from: tailcalled
comment by tailcalled · 2024-01-03T22:25:25.192Z · LW(p) · GW(p)

I've been meaning to write it up but I haven't gotten around to it yet, sorry.