Ahhh, but it is not clear to me that this is that disproportionate. In particular, I think this is a problem of EA people having more positive priors about MacAskill. Guzey then starts with more neutral priors, and then correctly updates downwards with his review, and then even more downwards when a promise of confidentiality was breached.
With regards to the contents of the book, I think the size of the downward updates exhibited in the essay dramatically exceeds the actual badness of what was found. Identifying errors is only the first step in an exercise like this – you then have to accurately update based on what those errors tell you. I think e.g. David Roodman’s discussion of this here is a much better example of the kind of work we want to see more of on the Forum.
With regards to the confidentiality screw-up, sure, it’s rational to update downwards in some general sense, but given that the actual consequences were so minor and that the alternative hypothesis (that it was just a mistake) is so plausible, I don’t respect Guzey’s presentation of this incident in his more recent writings (e.g. here).
Do you believe that the following representation of the incident is unfair?
Yes, at present I do.
I haven’t yet seen evidence to support the strong claims you are making about Julia Wise’s knowledge and intentions at various stages in this process. If your depiction of events is true (i.e. Wise both knowingly concealed the leak from you after realising what had happened, and explicitly lied about it somewhere) that seems very bad, but I haven’t seen evidence for that. Her own explanation of what happened seems quite plausible to me.
(Conversely, we do have evidence that MacAskill read your draft, and realised it was confidential, but didn’t tell you he’d seen it. That does seem bad to me, but much less bad than the leak itself – and Will has apologised for it pretty thoroughly.)
Your initial response to Julia’s apology seemed quite reasonable, so I was surprised to see you revert so strongly in your LessWrong comment a few months back. What new evidence did you get that hardened your views here so much?
And that since “the actual consequences were so minor and that the alternative hypothesis (that it was just a mistake) is so plausible” this doesn’t really matter?
It matters – it was a serious error and breach of Wise’s duty of confidentiality, and she has acknowledged it as such (it is now listed on CEA’s mistakes page). But I do think it is important to point out that, other than having your expectation of confidentiality breached per se, nothing bad happened to you.
One reason I think this is important is because it makes the strong “conspiracy” interpretation of these events much less plausible. You present these events as though the intent of these actions was to in some way undermine or discredit your criticisms (you’ve used the word “sabotage”) in order to protect MacAskill’s reputation. But nobody did this, and it’s not clear to me what they plausibly could have done – so what’s the motive?
What sharing the draft with MacAskill did enable was a prepared response – but that’s normal in EA and generally considered good practice when posting public criticism. Said norm is likely a big part of the reason this screw-up happened.
Ahhh, but it is not clear to me that this is that disproportionate. In particular, I think this is a problem of EA people having more positive priors about MacAskill. Guzey then starts with more neutral priors, and then correctly updates downwards with his review, and then even more downwards when a promise of confidentiality was breached.
Am I missing something here?
With regards to the contents of the book, I think the size of the downward updates exhibited in the essay dramatically exceeds the actual badness of what was found. Identifying errors is only the first step in an exercise like this – you then have to accurately update based on what those errors tell you. I think e.g. David Roodman’s discussion of this here is a much better example of the kind of work we want to see more of on the Forum.
With regards to the confidentiality screw-up, sure, it’s rational to update downwards in some general sense, but given that the actual consequences were so minor and that the alternative hypothesis (that it was just a mistake) is so plausible, I don’t respect Guzey’s presentation of this incident in his more recent writings (e.g. here).
(deleted)
Yes, at present I do.
I haven’t yet seen evidence to support the strong claims you are making about Julia Wise’s knowledge and intentions at various stages in this process. If your depiction of events is true (i.e. Wise both knowingly concealed the leak from you after realising what had happened, and explicitly lied about it somewhere) that seems very bad, but I haven’t seen evidence for that. Her own explanation of what happened seems quite plausible to me.
(Conversely, we do have evidence that MacAskill read your draft, and realised it was confidential, but didn’t tell you he’d seen it. That does seem bad to me, but much less bad than the leak itself – and Will has apologised for it pretty thoroughly.)
Your initial response to Julia’s apology seemed quite reasonable, so I was surprised to see you revert so strongly in your LessWrong comment a few months back. What new evidence did you get that hardened your views here so much?
It matters – it was a serious error and breach of Wise’s duty of confidentiality, and she has acknowledged it as such (it is now listed on CEA’s mistakes page). But I do think it is important to point out that, other than having your expectation of confidentiality breached per se, nothing bad happened to you.
One reason I think this is important is because it makes the strong “conspiracy” interpretation of these events much less plausible. You present these events as though the intent of these actions was to in some way undermine or discredit your criticisms (you’ve used the word “sabotage”) in order to protect MacAskill’s reputation. But nobody did this, and it’s not clear to me what they plausibly could have done – so what’s the motive?
What sharing the draft with MacAskill did enable was a prepared response – but that’s normal in EA and generally considered good practice when posting public criticism. Said norm is likely a big part of the reason this screw-up happened.
(deleted)