Sam Altman Q&A Notes - Aftermath

post by p.b. · 2021-09-08T08:20:56.702Z · LW · GW · 35 comments

To me it was quite painful when I found out after the first one of Sam Altman's Q&As that there was no recording and no way for me to find out what had been said. Which is why this time around I took notes and shared them here to help out people in the same situation.

This was initially very well received, although in the first 24+ hours a discussion arose, whether the notes were accurately representing what Sam Altman had said. I clarified some of the notes after feedback from other participants, and while not everybody was happy, I think that they were accurate within the limits of a very short summary. This was corroborated by other participants of the meetup.

Significantly later a different discussion started, about whether publishing these notes was some kind of breach of trust, because Sam Altman had requested for the meeting not to be recorded. 

To me avoiding a recording was about staying out of the mainstream media and not getting pinned down on concrete statements about sensitive topics. (One of his answers started with "Oh, this is a dangerous question ...", I did not include this in the notes.)

To my mind it was about being able "to speak more freely" as opposed to being able to speak at all. I did not believe that on the one hand he wanted to provide this information publicly, but on the other hand did not want to see it discussed.

I did not see this meetup as clandestine meeting of the alignment elite that is allowed to know about these matters, but possibly, I was wrong. (Ah, allow me this bit of sarcasm. Just this little bit. I already cut all? the bitterness.)

Some people argued that this was obviously a "defection" or at least a "misunderstanding of the rules" on my part. To which I can only point to the invitation [? · GW] which mentions neither recording, nor transcript, nor notes. In the meetup itself there was no mention of any prohibition to publicly discuss the provided information in writing or otherwise.

More sensibly, to my mind, some people argued that we shouldn't risk not having any more meetups with Sam Altman in the future. In the end moderators decided to take down the post until feedback by Sam Altman. A decision that I very much support. 

UPDATE: 

Sam Altman has now replied (see comments) that he likes to be able to interact with this community in an "off the record-ish way", so my notes will stay down. 

To me this sounds like he also doesn't have an exact clearance level in mind, but prefers to keep it low key. His feedback seems to have been really positive so hopefully there will be further Q&As in the future. 

There are probably many who couldn't attend the meetup due to real life interference, unfortunate timezones, because they missed the anouncement, or due to the fact that capacity was limited to 250 attendees.

So if you want me to contribute to a private Google doc, for instance, just send me a message or write a comment (though it might take longer for me to notice the comment).

35 comments

Comments sorted by top scores.

comment by Bjartur Tómas · 2021-09-08T21:08:24.324Z · LW(p) · GW(p)

Though my preferred outcome would be you taking the post down without much of a fuss, I understand this is a pretty self-serving preference. I did like the compromise idea of making the post available only to members, but that does not appear to be an existing feature of the site. 

Taking it down helps remedy a failure of my own rather than yours, as we clearly should have been more explicit about this. 

You posting them initially is perfectly understandable. Though I disagreed with your desire to keep them up after I requested them down, I understand this is a matter of opinion.

"Defection" is a pretty loaded word and I should not have used it. 

In general, I think it is really great when people provide public goods like book reviews or highlights (I also think it is really rewarding and I have never regretted doing such things myself), so to the extent this has discouraged you from this path, I would like to point out that this is obviously a weird “scissor case” and similar efforts in the future will certainly be well received. 

Replies from: habryka4
comment by habryka (habryka4) · 2021-09-08T23:49:51.972Z · LW(p) · GW(p)

The moderators were also discussing that it would have been nice to make the post only accessible to registered users, or users above a certain karma threshold, but yeah, it's not a feature that exists, and given the urgency of the question (given that there was a substantial amount of externally linked traffic coming in), it didn't seem like a good idea to wait until we built the feature.

comment by Ruby · 2021-09-08T20:52:27.481Z · LW(p) · GW(p)

I’m the moderator with final say on this matter, so I’ll lay out elements of  my current thinking:

  1. p.b. did a very reasonable and prosocial thing by taking notes and sharing them. It sounds like the confidentiality request was ambiguous on the topic of notes, and p.b.’s reasoning about it feels reasonable to me. I was compelled by it.
  2. It is generally good and cooperative to respect requests for confidentiality. In this case, it seems one could have either interpreted Sam’s request as definitely including written notes, or been unsure about his request but preferred erring on the side of caution. I think those are also reasonable positions.
  3. The situation/confidentiality request was ambiguous. On the one hand, a request was made. On the other, this was a talk given to a large number of people and the request wasn’t made particularly hard, e.g., not being made in advance, not being made in writing, etc.
  4. LessWrong should have a high bar for removing any user’s content. The benefits of information being shared by default are large and excessive caution in cases of ambiguity would both increase cognitive overhead and frequently result in important information not reaching the people who would benefit a lot from it.
  5. Sam’s preference on the matter is relevant. If he’s cool with the post, then no problem. If he prefers it to be removed, that factors into both what the mods and the author decide to do.
  6. I feel sympathetic to the ACX organizers who hosted Sam who fear that a violation of confidentiality request is a disincentive for future speakers (Sam and others). More generally, because I value information being shared, I’m afraid of things that would reduce such as people not feeling comfortable speaking.
  7. To the extent someone doesn’t want attention drawn to information, it can be better to not remove it once posted. And honoring Sam’s preferences at this point might actually mean leaving it up.

Last night, we pinged someone close to Sam about whether he’d prefer the post be taken down. They said that with high confidence they expect him to prefer that. If we hadn’t received this response, we would very likely have left the post up, since I don’t think preemptively taking it down based on an unclear confidentiality request would have been appropriate. We have also pinged Sam himself but haven’t heard back yet.

My current leaning is that if p.b. wants to leave it down, that’s definitely their prerogative. If Sam responds and wants us to keep it down and this seems like a reasonable request backed by compelling reasons, we’ll likely honor that. Correspondingly, if there aren’t compelling reasons, e.g., motivations driven by something deceptive, then we would encourage p.b. to republish it. In the absence of a clear response, we’ll leave the decision up to p.b. after sharing with them all the things we know.

Replies from: Ruby, Davidmanheim
comment by Ruby · 2021-09-10T01:03:50.295Z · LW(p) · GW(p)

Sam has replied to us:

  • He really appreciates us taking it down since he doesn't usually talk in much detail about future plans.
  • He really enjoys engaging with this community and it's nice to able to do this in an off-the-record-ish way.

That seems reasonable to me. If I had to guess, many people are very interested in Sam's plans which makes them hard to talk about publicly (people reading too much into every little thing, misinterpreting things, etc, and he probably doesn't want people reacting to misunderstandings, etc.). Given that, I think the content should stay down out of respect for Sam's wishes.  

Replies from: RobbBB
comment by Rob Bensinger (RobbBB) · 2021-09-10T01:17:59.698Z · LW(p) · GW(p)

Good to know!

Sam apparently said at the meetup that it was fine to share notes with friends, so I assume it's fine to pass around private copies of p.b.'s notes (and Gwern's LW comments, etc.) in a Google Doc -- manually adding each individual who wants to see it, rather than setting it to 'anyone with the link can view'.

Replies from: RobbBB
comment by Rob Bensinger (RobbBB) · 2021-09-10T01:31:50.015Z · LW(p) · GW(p)

There's also an ambiguity here about "how much can LWers discuss particular important questions or issues that are informed by stuff they learned from the meetup, or from p.b. and others' notes?".

I assume Sam's views are now generally 'in the water' on LW, and that it would be pretty costly for the whole community to permanently self-censor about them in all non-private discussion.

(Both because the info is pretty important and discussion-worthy, and because keeping long-term secrets tends to be very cognitively taxing. And people weren't warned in advance that they might have to permanently tag a bunch of complicated facts in their head as 'secret' and avoid mentioning them at all -- either at the meetup, AFAICT, or before reading p.b.'s post.)

I think it makes more sense for the norm to be something like 'don't make a big public list of Things Sam Said At The Meetup', 'don't go giving interviews to reporters about it', and 'don't title your posts "Sam Altman thinks X" if your only evidence is from the meetup', but for discussion of this stuff to otherwise be fair game going forward. (Maybe after a one-week moratorium or something, I dunno.)

Replies from: Dagon, RobbBB
comment by Dagon · 2021-09-10T16:55:28.621Z · LW(p) · GW(p)

+1
My model of this (was not at the meetup, but have been part of many discussions/announcements with varying levels of confidentiality) is that talking about issues or topics is just fine, but quotes or appeals to authority "Sam Altman said!" are discouraged.  Anything Privileged and Confidential (legally non-discoverable) was already not part of the Q&A.  Anything so tentative or sensitive that it shouldn't be discussed at all will have been explicitly stated in the meeting as such (or omitted entirely).

comment by Rob Bensinger (RobbBB) · 2021-09-10T01:34:00.289Z · LW(p) · GW(p)

(If others disagree, I'm probably easy to convince on this.)

comment by Davidmanheim · 2021-09-09T04:30:57.234Z · LW(p) · GW(p)

The request for this to be off the record was explicit during the introduction to the talk, so I'm not sure why it's ambiguous. And "off the record" has a pretty clear meaning - I certainly had the clear expectation that my question, and his answer, weren't going to be published.

Edit: I do not recall the phrasing, but as I said below, I was under a distinct impression that the request for no recording and no transcript was at least indicative, and that asking him if you could share notes publicly would have been the right thing to do.

Replies from: p.b., Ruby
comment by p.b. · 2021-09-09T16:02:14.488Z · LW(p) · GW(p)

I definitely don't remember the terms "off the record" being used. 

And I think if the other participants of the meetup who commented on the post, had any memory of these terms being used, they would have mentioned it. Because, yes, that's not very ambiguous and I don't think there would have been much of a discussion then. 

Replies from: habryka4, Davidmanheim
comment by habryka (habryka4) · 2021-09-09T19:25:44.707Z · LW(p) · GW(p)

Whether this term was being used also feels pretty cruxy to me, so other people chiming in would be useful (ironically it would be useful to have a recording of the talk so we could figure out what confidentiality request was made :P). 

comment by Davidmanheim · 2021-09-11T17:45:32.015Z · LW(p) · GW(p)

I don't have a very good (or even halfway decent) memory for phrases, so I have no idea, and since no-one else heard it, I assume it wasn't said. Still, it seemed clear to me that the request was intended for the talk to be off the record, in the journalistic sense. 

The phrase "no recording and no transcript," which you seem to agree was said explicitly, seems to indicate that he didn't want there to be a record of what he said. At that point,  maybe you didn't technically do anything he requested you not to do, but it seems like the responsible  and decent thing would be to have asked Sam if he minded.

comment by Ruby · 2021-09-09T05:27:19.141Z · LW(p) · GW(p)

If the words "off the record" were used, that does feel stronger. I wasn't there. My understanding was that "no recording and no transcript" was what was requested.

Replies from: p.b., Davidmanheim
comment by p.b. · 2021-09-09T16:04:25.646Z · LW(p) · GW(p)

It may have been requested from the organizers and it may have been mentioned that there won't be a recording, but as far as I remember it was not requested from the participants.

comment by Davidmanheim · 2021-09-11T17:30:46.691Z · LW(p) · GW(p)

I don't recall what phrase was used, but I thought that it was clear enough. If someone said that they agree to do a talk on the condition that there be no recording and no transcript, unlike every other talk in the series, it seems to take a really weird model of the situation to claim that you had no idea that they would not want people publicly posting notes. At the very least, it merits checking.

Replies from: Ruby
comment by Ruby · 2021-09-11T17:50:17.215Z · LW(p) · GW(p)

I think there's a difference between "recording and transcript" and "some guy's rough notes". If the concern is journalist-quote-mining, the latter might be substantially less useful to a journalist while still be useful to people in the community.

I agree that checking would have been a good idea, but notes definitely feel they like they fall in ambiguous zone given what was said according to people's reports.

comment by Vanessa Kosoy (vanessa-kosoy) · 2021-09-08T14:47:13.012Z · LW(p) · GW(p)

Just wanted to say that you did nothing wrong IMO. Also, I feel like I got some benefit from the notes, and the accuracy criticism seemed so weak that the notes were probably fairly accurate.

comment by jimrandomh · 2021-09-08T17:56:43.107Z · LW(p) · GW(p)

This was a significant controversy among the moderators. I believed, and continue to believe, that the post should have stayed up.

Replies from: Davidmanheim
comment by Davidmanheim · 2021-09-09T04:28:19.251Z · LW(p) · GW(p)

My simple question is whether he checked with Sam Altman and/or Joshua Fox about whether posting notes of this explicitly off the record talk, as mentioned during the talk, was OK before posting - and as far as I can tell, he did not.

Replies from: stella-biderman
comment by Stella Biderman (stella-biderman) · 2021-09-09T22:55:36.381Z · LW(p) · GW(p)

When was it stated that the talk was off the record? You seem to be the only person in this thread (myself included) who remembers that.

Replies from: Davidmanheim
comment by Davidmanheim · 2021-09-11T17:26:04.198Z · LW(p) · GW(p)

Sam has said he thought this was "off-the-record-ish" and it was clearly known that it not being recorded was a precondition for giving the talk. I don't recall what terms were used, but I thought it was pretty obvious - and Sam's later responses seem to agree - that he expected notes like this not to be made public.

Edit to add: I thought at the time that it was clear that this was off the record, despite that phrase likely not being used. If not, I would not have asked the question which I asked during the meetup.

Replies from: Ruby
comment by Ruby · 2021-09-11T17:57:27.794Z · LW(p) · GW(p)

Sam's reply indicated that his preference was for it to be taken down. I wouldn't interpret that as him expressing a violated expectation of what would happen given what was said. 

Plus "off-the-recordish-ish", would imply he doesn't expect the information to be perfectly siloed.  What does "ish" mean exactly? Unclear, not that that phrase was used at the event anyhow. Overall, seems it was ambiguous and the notes were an edge case.

The thing about situations of ambiguity is things that feel obvious to some people don't feel obvious to everyone. I think I personally would have erred more on the side of caution like you, but I don't think p.b. did something super obviously wrong. I don't think Sam's preferences were super well specified.

Replies from: Davidmanheim
comment by Davidmanheim · 2021-09-11T19:59:31.434Z · LW(p) · GW(p)

That's all fair, and given what has been said, despite my initial impression, I don't think this was "obviously wrong" - but I do have a hope that  in this community, especially in acknowledged edge cases, people wait and check with others rather than going ahead.

Replies from: Ruby
comment by Ruby · 2021-09-11T22:35:50.069Z · LW(p) · GW(p)

Maybe. You can be too biased in either direction. One direction and you violate privacy which makes people not say things, in the other direction people are afraid of violating privacy and therefore valuable information doesn't get spread (because asking is effortful, or highly impractical, e.g. Sam Altman isn't that accessible). People should use their judgment, and sometimes they'll get it wrong.

Replies from: Davidmanheim
comment by Davidmanheim · 2021-09-13T08:39:11.521Z · LW(p) · GW(p)

I agree that in general there is a tradeoff, and that there will always be edge cases. But in this case, I think judgement should be tilted strongly in favor of discretion. That's because a high trust environment is characterized by people being more cautious around public disclosure and openness. Similarly, low trust environments have higher costs of communication internal to the community, due to lack of willingness to interact or share information. Given the domain discussed, and the importance of collaboration between key actors in AI safety, I'm by default in favor of putting value more on higher trust and less disclosure than on higher transparency and more sharing.

comment by James_Miller · 2021-09-08T16:10:07.017Z · LW(p) · GW(p)

While the decision to remove might have been correct, clearly Altman placed some value on people like us learning about his views and might have appreciated your efforts to inform more such people about this thinking in a way that's opaque to the traditional media.

comment by JenniferRM · 2021-09-12T05:10:00.571Z · LW(p) · GW(p)

I don't think Lesswrong should censor things that aren't obviously just spammy ads or botnets or such.

I don't think making Sam Altman "more comfortable [LW · GW]" is a valid reason to censor things. 

If he didn't want someone to take notes and spread the good ideas in them, he shouldn't have been giving a speech to thoughtful and kind and morally scrupulous people.

Certainly, if there is a call to be made, the call should be made by the author, not the mods.

comment by Tom (tom-1) · 2021-09-08T19:06:47.166Z · LW(p) · GW(p)

I'm kinda bummed. I got half way through and was finding it fascinating, but then got interrupted. When I tried to return it was gone.

I will subscribe to this page in the hopes that I hear of the return of the notes.

comment by Gunnar_Zarncke · 2021-09-08T18:29:57.671Z · LW(p) · GW(p)

I can very much relate to the painful feeling of having a post pulled. It has happened to me too. On the other hand, this seems to be a transient state and it is not unlikely that Sam Altman will approve of it. And if he doesn't, well. I guess that's what you meant with "A decision that I very much support.".

Replies from: JenniferRM
comment by JenniferRM · 2021-09-12T05:13:37.815Z · LW(p) · GW(p)

I'm curious what post was pulled. I had thought that Lesswrong was essentially free from censorship, but now I'm updating in the direction that it was just very effective at censorship, and maybe is therefore not a safe place to hang out.

Do you still endorse the thing that was pulled? Can you link to it here, or PM it to me?

Replies from: Gunnar_Zarncke, tom-1
comment by Gunnar_Zarncke · 2021-09-12T09:43:42.486Z · LW(p) · GW(p)

Well, spam is deleted all the time. But I have never seen an article pulled completely before. In this regard what happened to my posts was different: They were demoted from main to discussions (that was before LW 2.0 when moderation worked differently). I was mostly relating to the feeling before posting.

Replies from: Vaniver, JenniferRM
comment by Vaniver · 2021-09-14T04:46:16.347Z · LW(p) · GW(p)

But I have never seen an article pulled completely before.

It happened before, but it's quite rare. Normally when I've done it, I've left a note in an Open Thread, such as this case [LW(p) · GW(p)] where I moved to drafts a post that was talking about an ongoing legal case (now concluded). I think that's the last one I did, and it was four years ago? But there are other mods as well.

comment by JenniferRM · 2021-09-13T01:51:27.063Z · LW(p) · GW(p)

This was new to me too, and I didn't like how muddy it made the waters. Thank you for the information! It bounds the badness, which is comforting <3

comment by Tom (tom-1) · 2021-09-12T22:09:55.244Z · LW(p) · GW(p)

I guess it depends on how you define censorship.  In this case, it wasn't pulled because the audience found it objectionable or because the owners of the medium wanted to silence the author or material. It was pulled, with the agreement of the author, because of an issue with a source.  

I don't consider it censorship if a newspaper and journalist agree to spike a story because of a legitimate issue with a source.

Replies from: JenniferRM
comment by JenniferRM · 2021-09-13T01:49:44.723Z · LW(p) · GW(p)

Newspapers are mostly garbage, and I do not, by default, admire their epistemic or typical "moral" standards.

To the degree that the first amendment is wise, it is by making sure that if evil powers are temporarily in control of the government, even then they can't prevent the spread of the truth, because the tools for the spreading of truth are so clearly protected that even the spreaders of intellectual sewage are granted the right to operate.

Newspapers rake muck through exaggeration, and spread falsehoods, and bash people indirectly during political seasons, and build viewership via tabloid antics when no elections are at stake.

The whole thing is is essentially patrimonial with the Owner/CEOs and the CEO's delegated Editors as the monarch-like rulers-over-the-mere-writers, with the powerful folk existing in a murky world of favors and dirty deals, and the minions towing the line as they must, because they are working stiffs who need a job, and some things are "above their pay grade". This works for some things, like making ice cream and building houses. Capitalism, baby!

Lesswrong does not pay the writers that I've heard of? And Lesswrong doesn't direct our writing. And I think probably: thank goodness for that?

You say:

I don't consider it censorship if a newspaper and journalist agree to spike a story

Contrast this with my claim over in a cousin comment (which did not have the bold over there):

Certainly, if there is a call to be made, the call should be made by the author, not the mods.

I don't want to harp too much on this bit, which feels close to the crux for me... 

Reading above, Ruby claims to have the final say... and used her final say to say that it was the author's say... 

This is a beautiful waffle, and Ruby perhaps deserves a very prestigious and beautiful Reagan Clinton Trump Teflon Waffleiron Award for wiggling out of this mess without ceding power or offending very many people <3

But like: If Lesswrong was just definitely and clearly a patrimonial system, and the mods often spiked stories as a favor to rich guys, because maybe the mods want to make the rich guy comfortable, because the rich guys have power and could do favors in return...

...then I think maybe I would be done with Lesswrong?

I wrote a lot more words about the abstract principle here, but I think they can be boiled down to:

(1) Sending written things down the memory hole is a bad policy for a platform that isn't just for commerce and shilling.

(2) I'm an inclusionist. Good people, cooperating to build pro-social communications infrastructure, should treat the censorship of truth as damage, and route around this damage.

(3) Lesswrong should not use people who aren't actually that good (like most for-profit news organizations and those who work in them) as role models.