I’m not asking you to engage with Mikhail more, I believe I understand it’s frustrating given your extensive prior conversations that still led to this post being made.
Nevertheless, I have found all these comments informative as well as op.
The post says Mikhail sent
Also, apparently, just in case, please don’t act on me having told you that [third party] are planning to do [thing] outside of this enabling you to chat to/coordinate with them)
and that you replied “lol, no” after a week.
I generally don’t want to clash with you as I respect a lot of your public takes etc, but for the same reasons you’re publicly disagreeable I do think it’s worth pointing my disagreement here. Unless you were already on colloquial terms with Mikhail, I find it rude you’d answer “lol no” to that specific request, notably given it used “please”. Even if it was an unreasonable ask, a “sorry but no” would have sufficed.
At the object level, as board member of enais and french centre for ai safety, I don’t even take Mikhail’s message as a surprising or unreasonable ask, unless interpreted stringently. Ofc if the formulation was “please make sure to act indistinguishably, even when assessed by a future superintelligence, on this info”, then a lol no is fine, but if Mikhail sent me this message I’d interpret it as asking the 80⁄20 reasonable effort and say I broadly agree tho will take other stuff into account too. In fact I overall guess (or want to believe) that you in fact do broadly agree (you prefer for people to tell you stuff, and soft commit to not indirectly harming people who tell you stuff if you can avoid it) and that you and Mikhail disagree because you’re interpreting Mikhail to be dogmatic and strict about his ask (which might actually be the case).
As an interested third party who generally would like to to work with LightConeInfra and you, unrelated to Mikhail’s specific asks, I’m curious for if you broadly agree to put some non trivial decision weight on not using info people give you in ways they strongly disagree with, even if they didn’t ask you to precomit to that, even if they were mistaken in some assumptions. (If you later get that info from other places you’re ~released from the first obligations, tho this shouldn’t be gamed)
No, what I did is reply with “lol, no” followed by about 3000 words of explanation across a 1-2 hour chat conversation, detailing my decision procedures, and what I am and am not happy to do. Like, I really went into a huge amount of detail, gave concrete specific examples, and elaborated what I would do. Much of this involved Mikhail insisting on a very specific interpretation of what reasonable conduct is and clarifying multiple times that yes, he wouldn’t want me to use information like this under any circumstance in any kind of way adversarial to the third party the information is about, and that it would be unreasonable for me to reject such a request.
As an interested third party who generally would like to to work with LightConeInfra and you, unrelated to Mikhail’s specific asks, I’m curious for if you broadly agree to put some non trivial decision weight on not using info people give you in ways they strongly disagree with, even if they didn’t ask you to precomit to that, even if they were mistaken in some assumptions. (If you later get that info from other places you’re ~released from the first obligations, tho this shouldn’t be gamed)
Of course! See my general process described above. If you tell me something in secret, or ask me to put some kind of constraint on information, I will check whether I would have accepted that information with that constraint in advance. If I would have, I am happy to agree to it afterwards. Similarly, if I think you have some important preference, but you just forgot to ask me explicitly, or we didn’t have time to discuss it, or it’s just kind of obvious that you have this preference, I will do the same.
I have a bunch more thoughts, but I don’t super want to prop up this comment section by writing stuff that I actually think is worth reading in general. I’ll post my more cleaned-up thoughts somewhere else and link them.
I won’t go public with it, but of course I am going to consider it in my plans in various ways
You can ask in-advance if I want to accept confidentiality on something, and I’ll usually say no
Many words followed only after I expressed surprise and started the discussion (“about 3000 words of explanation, and a 1-2 hour chat conversation” is false, there were fewer than 2k words from your side in the entire conversation, many of which were about the third party, unrelated to your explanations of your decision procedures etc., a discussion of making a bet that you ended up not taking, some facts that you got wrong, etc.)
insisting again and again on a very specific interpretation of what reasonable conduct is and clarifying multiple times that yes, he wouldn’t want me to use information like this under any circumstance in any kind of way adversarial to the third party the information is about
I think you’re misrepresenting what I asked; I asked you to not use it adversarially towards the third party, as it seemed to me as a less strong demand than confidentiality, especially given that you already shared it with people and also said you want to be able to share information and think people like Eliezer are wrong about all the keeping-secrets stuff.
If you tell me something in secret, or ask me to put some kind of constraint on information, I will check whether I would have accepted that information with that constraint in advance. If I would have, I am happy to agree to it afterwards
Similarly, if I think you have some important preference, but you just forgot to ask me explicitly, or we didn’t have time to discuss it, or it’s just kind of obvious that you have this preference, I will do the same
Among the almost 2000 words, you did not describe this procedure even once.
And maybe I completely misinterpreted what you wrote, but from your messages, I had the impression of quite the opposite: that you think it is insane to expect people to use information in ways that align with important preferences.
Many words followed only after I expressed surprise and started the discussion (“about 3000 words of explanation, and a 1-2 hour chat conversation” is false, there were fewer than 2k words from your side in the entire conversation, many of which were about the third party, unrelated to your explanations of your decision procedures etc., a discussion of making a bet that you ended up not taking, some facts that you got wrong, etc.)
When I just extracted my messages from the thread I was referencing and threw them into a wordcounter I got 2,370 words in my part of the conversation (across three parallel threads), which is close enough to 3,000 that I feel good about my estimate. I do now realize that about 500 of those were a few weeks later (but still like a month ago), so I would have now said more like 2000 words to refer to that specific 1-2 hour conversation (do appreciate the correction, though I think in this context the conversation a few weeks later makes sense to include).
Among the almost 2000 words, you did not describe this procedure even once.
I brought it up as a consideration a few times. (Example: “Like, to be clear, I definitely rather you not have told me instead of demanding that [I] ‘only use the information to coordinate’ afterwards”). I agree I didn’t outline my whole decision-making procedure, but I did explain .
that you think it is insane to expect people to use information in ways that align with important preferences.
Sorry, I am not parsing this. My guess is you meant to say something else than “important preferences” here?
I think you’re misrepresenting what I asked; I asked you to not use it adversarially towards the third party, as it seemed to me as a less strong demand than confidentiality
It’s plausible I am still not understanding what you are asking. To be clear, what you asked for seemed to me substantially costlier than confidentiality (as I communicated pretty early on after you made your request). I have hopefully clarified my policies sufficiently now.
This kind of stuff is hard and we are evidently not on the same page about many of the basics, and that’s part of why I don’t feel comfortable promising things here, since my feeling is that you feel pretty upset already about me having violated something you consider an important norm, and I would like to calibrate expectations.
I’m not asking you to engage with Mikhail more, I believe I understand it’s frustrating given your extensive prior conversations that still led to this post being made.
Nevertheless, I have found all these comments informative as well as op.
The post says Mikhail sent
and that you replied “lol, no” after a week.
I generally don’t want to clash with you as I respect a lot of your public takes etc, but for the same reasons you’re publicly disagreeable I do think it’s worth pointing my disagreement here. Unless you were already on colloquial terms with Mikhail, I find it rude you’d answer “lol no” to that specific request, notably given it used “please”. Even if it was an unreasonable ask, a “sorry but no” would have sufficed.
At the object level, as board member of enais and french centre for ai safety, I don’t even take Mikhail’s message as a surprising or unreasonable ask, unless interpreted stringently. Ofc if the formulation was “please make sure to act indistinguishably, even when assessed by a future superintelligence, on this info”, then a lol no is fine, but if Mikhail sent me this message I’d interpret it as asking the 80⁄20 reasonable effort and say I broadly agree tho will take other stuff into account too. In fact I overall guess (or want to believe) that you in fact do broadly agree (you prefer for people to tell you stuff, and soft commit to not indirectly harming people who tell you stuff if you can avoid it) and that you and Mikhail disagree because you’re interpreting Mikhail to be dogmatic and strict about his ask (which might actually be the case).
As an interested third party who generally would like to to work with LightConeInfra and you, unrelated to Mikhail’s specific asks, I’m curious for if you broadly agree to put some non trivial decision weight on not using info people give you in ways they strongly disagree with, even if they didn’t ask you to precomit to that, even if they were mistaken in some assumptions. (If you later get that info from other places you’re ~released from the first obligations, tho this shouldn’t be gamed)
No, what I did is reply with “lol, no” followed by about 3000 words of explanation across a 1-2 hour chat conversation, detailing my decision procedures, and what I am and am not happy to do. Like, I really went into a huge amount of detail, gave concrete specific examples, and elaborated what I would do. Much of this involved Mikhail insisting on a very specific interpretation of what reasonable conduct is and clarifying multiple times that yes, he wouldn’t want me to use information like this under any circumstance in any kind of way adversarial to the third party the information is about, and that it would be unreasonable for me to reject such a request.
Of course! See my general process described above. If you tell me something in secret, or ask me to put some kind of constraint on information, I will check whether I would have accepted that information with that constraint in advance. If I would have, I am happy to agree to it afterwards. Similarly, if I think you have some important preference, but you just forgot to ask me explicitly, or we didn’t have time to discuss it, or it’s just kind of obvious that you have this preference, I will do the same.
I have a bunch more thoughts, but I don’t super want to prop up this comment section by writing stuff that I actually think is worth reading in general. I’ll post my more cleaned-up thoughts somewhere else and link them.
Your entire reply, after a week, was:
Many words followed only after I expressed surprise and started the discussion (“about 3000 words of explanation, and a 1-2 hour chat conversation” is false, there were fewer than 2k words from your side in the entire conversation, many of which were about the third party, unrelated to your explanations of your decision procedures etc., a discussion of making a bet that you ended up not taking, some facts that you got wrong, etc.)
I think you’re misrepresenting what I asked; I asked you to not use it adversarially towards the third party, as it seemed to me as a less strong demand than confidentiality, especially given that you already shared it with people and also said you want to be able to share information and think people like Eliezer are wrong about all the keeping-secrets stuff.
Among the almost 2000 words, you did not describe this procedure even once.
And maybe I completely misinterpreted what you wrote, but from your messages, I had the impression of quite the opposite: that you think it is insane to expect people to use information in ways that align with important preferences.
When I just extracted my messages from the thread I was referencing and threw them into a wordcounter I got 2,370 words in my part of the conversation (across three parallel threads), which is close enough to 3,000 that I feel good about my estimate. I do now realize that about 500 of those were a few weeks later (but still like a month ago), so I would have now said more like 2000 words to refer to that specific 1-2 hour conversation (do appreciate the correction, though I think in this context the conversation a few weeks later makes sense to include).
I brought it up as a consideration a few times. (Example: “Like, to be clear, I definitely rather you not have told me instead of demanding that [I] ‘only use the information to coordinate’ afterwards”). I agree I didn’t outline my whole decision-making procedure, but I did explain .
Sorry, I am not parsing this. My guess is you meant to say something else than “important preferences” here?
It’s plausible I am still not understanding what you are asking. To be clear, what you asked for seemed to me substantially costlier than confidentiality (as I communicated pretty early on after you made your request). I have hopefully clarified my policies sufficiently now.
This kind of stuff is hard and we are evidently not on the same page about many of the basics, and that’s part of why I don’t feel comfortable promising things here, since my feeling is that you feel pretty upset already about me having violated something you consider an important norm, and I would like to calibrate expectations.