I tried to look at Daniel Filan’s tweets. The roughly 240 posts long megathread is a largely unstructed stream of consciousness that appears hard to read. He doesn’t seem to make direct arguments so much as it is a scratchpad for thoughts and questions that occured to the author while listening to the debate. The mere act of pointing to such a borderline unreadable thread IMHO doesn’t itself constitute a significant counterargument, nor does the act of referring to the prospect of a future summary article. Reasonable norms of good debate suggest relevant counterarguments should be proportional in length and readability to the original argument, which in this case is Rokos compact nine-minute post.
Again, imagine me making a case for or against AI risk by pointing to a 17-hour YouTube debate and to an overly long and convoluted Twitter thread of the thoughts of some guy who listened to the debate. I think few would take me seriously.
I also note that you originally seemed to suggest you watched the debate in whole, while you now sound as if you watched only parts of it and read Daniel Filan’s thread. If this alone really got you “from about 70% likelihood of a lab-leak to about 1-5%”, then I think it should be easy for you to post an object-level counterargument to Roko’s post.
(Apart from that, Daniel Filan himself says he is 75-80% convinced of the zoonosis hypothesis, in contrast to your 95-99%, while also noting that this estimate doesn’t even include considerations about genetic evidence which he apparently expects to favor the lab-leak hypothesis. He also says: “Oh also this is influenced by the zoonosis guy being more impressive, which may or may not be a bias.” Though pointing to other people’s credences doesn’t provide a significant argument for anything unless there is also evidence this individual is unusually reliable at making such judgements.)
> Reasonable norms of good debate suggest relevant counterarguments should be proportional in length and readability to the original argument, which in this case is Rokos compact nine-minute post.
This seems entirely *un*reasonable to me. Some arguments simply can’t be properly made that concisely, and this principle seems to bias us towards finding snappy, simplistic explanations rather than true ones.
Someone else mentioned ‘The Pyramid and the Garden’, but I’m reminded of the sort of related argument about Atlantis in https://slatestarcodex.com/2016/11/16/you-are-still-crying-wolf/ : sometimes, boring reality and ‘it’s actually just a series of coincidences’ requires a lot more explaining than a neat little conspiracy theory. Not to tar the lab leak hypothesis by calling it a conspiracy theory- while of course it literally is one, it doesn’t deserve to be demeaned by the term’s modern connotation of zany insane-person nonsense- but it is easy to see why ‘these facts seems unlikely to be a coincidence’ might be easier to argue concisely than its rebuttal, and a norm where the person that can state their argument more persuasively in shortform wins doesn’t seem like one that’s going to promote optimal truth-seeking.
The problem with “lab-leak is unlikely, look at this 17-hour debate” is that it is too short an argument, not a too long one. Arguments don’t get substituted by merely referring to them. The expression “the Riemann hypothesis” is not synonymous with the Riemann hypothesis. The former just refers to the latter. You can understand one without understanding the other.
On average, every argument gets more indirect the less accessible it is, and inaccessibility is strongly dependent on length, as well as language, intelligibility, format etc.
It may be that the 17-hour debate cannot be summarized adequately (depending on some standard of adequacy) in a nine-minute post, but a nine-minute post would be much more adequate than “lab-leak is unlikely, look at this 17-hour debate”.
The problem with “lab-leak is unlikely, look at this 17-hour debate” is that it is too short an argument, not a too long one.
It isn’t an argument, it’s a citation.
I don’t think a 17 hour debate is “inaccessible” to someone who is invested in this issue and making extremely strong, potentially very seriously libellous claims without having investigated some of the central arguments on the question at hand.
A foundational text in some academic field might take 17 hours to read, but you would still expect someone to have read it before making a priori wild claims that contradicted the expert consensus of that field very radically. I don’t think you’d take that person seriously at all if they hadn’t, and would in fact consider it very irresponsible (and frankly idiotic) for them to even make the claims until they had.
That’s not to say that this debate should be treated as foundational to the study of this question, exactly, but… well, as I said elsewhere:
This debate has been cited repeatedly in rationalist spaces, by people who were already quite engaged with the topic, familiar with the evidence, and in possession of carefully-formed views, as having been extremely valuable and informative, and having shifted their position significantly.”
I think that makes familiarising yourself with those arguments (whether from the debate or another equivalent-or-better source) a prerequisite for making the kind of strong, confident claims Roko is making. At the moment, he’s making those claims without the information necessary to be anywhere near as confident as he is.
A foundational text in some academic field might take 17 hours to read, but you would still expect someone to have read it before making a priori wild claims that contradicted the expert consensus of that field very radically.
There’s a good reason why the foundational text in an academic field would be a text and not a video. Evaluating arguments is easier to do when they are done via text.
I think that makes familiarising yourself with those arguments (whether from the debate or another equivalent-or-better source)
I don’t think anyone has shown that the debate contains specific arguments that Roko is unfamiliar with.
Evaluating arguments is easier to do when they are done via text.
Isn’t there a transcript? In any case, this seems to be highly subjective, and in my opinion not hugely relevant anyway. To extend the analogy, your expectation of someone’s having read foundational texts before making strong claims would hardly be lessened by the objection that the texts were hard to read.
I don’t think anyone has shown that the debate contains specific arguments that Roko is unfamiliar with.
Do we need to positively show that? As I mentioned, many intelligent, thoughtful people who were already very familiar with this question and its relevant facts updated significantly based on the debate. And Roko certainly hasn’t explicitly addressed all of the arguments therein. Isn’t that enough to suggest that he should at least watch it?
It’s equivalent to watching a season of a TV show- is that really such an onerous requirement for making incredibly strong, potentially libellous claims about a contentious issue with serious real-world ramifications?
If you watch a debate the way you would a TV show where you aren’t critically evaluating everything, that’s not a good basis for forming beliefs about the real world.
Youtube videos tend to encourage you to consume them in that way, but that doesn’t make it a good epistemic practice and that’s part of the reason why scientific debates usually happen in text.
Yep, the tweet thread is just me jotting down thoughts etc while watching the debate.
I was 75-80% convinced when I tweeted that, which was before I had finished the debate. After watching the debate, I made a sketchy Bayesian calculation that got me to 96%, but I’ve since backed off to maybe 66%.
Basically: the key question for me is whether you think one of the first outbreaks of COVID happened at the Huanan Seafood Market. Rootclaim conceded this, and as far as I can tell if this is true then it’s dispositive evidence, but I have since began to doubt it.
Sure, my post doesn’t rebut Roko’s. It isn’t an arguement, after all. It is a recommendation for some content.
Yep, I didn’t watch the whole debate. I apologize if I gave that impression. Mea culpa.
How much you update on the evidence depends on your priors, no? Daniel went from “probably lab-leak” to “probably zoonosis”, and I went from “IDK” to “quite probably lab-leak”. That is the similairity I was pointint at.
I found Daniel’s thread helpful after watching some hours of the debate, as he noticed numerous points I missed. Plus, I had watched enough that the rest of the data in the thread made sense to me. So you’re right, Daniel’s thread probably isn’t that helpful. At least, not without watching e.g. the opening arguements by rootclaim and peter miller.
I tried to look at Daniel Filan’s tweets. The roughly 240 posts long megathread is a largely unstructed stream of consciousness that appears hard to read. He doesn’t seem to make direct arguments so much as it is a scratchpad for thoughts and questions that occured to the author while listening to the debate. The mere act of pointing to such a borderline unreadable thread IMHO doesn’t itself constitute a significant counterargument, nor does the act of referring to the prospect of a future summary article. Reasonable norms of good debate suggest relevant counterarguments should be proportional in length and readability to the original argument, which in this case is Rokos compact nine-minute post.
Again, imagine me making a case for or against AI risk by pointing to a 17-hour YouTube debate and to an overly long and convoluted Twitter thread of the thoughts of some guy who listened to the debate. I think few would take me seriously.
I also note that you originally seemed to suggest you watched the debate in whole, while you now sound as if you watched only parts of it and read Daniel Filan’s thread. If this alone really got you “from about 70% likelihood of a lab-leak to about 1-5%”, then I think it should be easy for you to post an object-level counterargument to Roko’s post.
(Apart from that, Daniel Filan himself says he is 75-80% convinced of the zoonosis hypothesis, in contrast to your 95-99%, while also noting that this estimate doesn’t even include considerations about genetic evidence which he apparently expects to favor the lab-leak hypothesis. He also says: “Oh also this is influenced by the zoonosis guy being more impressive, which may or may not be a bias.” Though pointing to other people’s credences doesn’t provide a significant argument for anything unless there is also evidence this individual is unusually reliable at making such judgements.)
> Reasonable norms of good debate suggest relevant counterarguments should be proportional in length and readability to the original argument, which in this case is Rokos compact nine-minute post.
This seems entirely *un*reasonable to me. Some arguments simply can’t be properly made that concisely, and this principle seems to bias us towards finding snappy, simplistic explanations rather than true ones.
Someone else mentioned ‘The Pyramid and the Garden’, but I’m reminded of the sort of related argument about Atlantis in https://slatestarcodex.com/2016/11/16/you-are-still-crying-wolf/ : sometimes, boring reality and ‘it’s actually just a series of coincidences’ requires a lot more explaining than a neat little conspiracy theory. Not to tar the lab leak hypothesis by calling it a conspiracy theory- while of course it literally is one, it doesn’t deserve to be demeaned by the term’s modern connotation of zany insane-person nonsense- but it is easy to see why ‘these facts seems unlikely to be a coincidence’ might be easier to argue concisely than its rebuttal, and a norm where the person that can state their argument more persuasively in shortform wins doesn’t seem like one that’s going to promote optimal truth-seeking.
The problem with “lab-leak is unlikely, look at this 17-hour debate” is that it is too short an argument, not a too long one. Arguments don’t get substituted by merely referring to them. The expression “the Riemann hypothesis” is not synonymous with the Riemann hypothesis. The former just refers to the latter. You can understand one without understanding the other.
On average, every argument gets more indirect the less accessible it is, and inaccessibility is strongly dependent on length, as well as language, intelligibility, format etc.
It may be that the 17-hour debate cannot be summarized adequately (depending on some standard of adequacy) in a nine-minute post, but a nine-minute post would be much more adequate than “lab-leak is unlikely, look at this 17-hour debate”.
It isn’t an argument, it’s a citation.
I don’t think a 17 hour debate is “inaccessible” to someone who is invested in this issue and making extremely strong, potentially very seriously libellous claims without having investigated some of the central arguments on the question at hand.
A foundational text in some academic field might take 17 hours to read, but you would still expect someone to have read it before making a priori wild claims that contradicted the expert consensus of that field very radically. I don’t think you’d take that person seriously at all if they hadn’t, and would in fact consider it very irresponsible (and frankly idiotic) for them to even make the claims until they had.
That’s not to say that this debate should be treated as foundational to the study of this question, exactly, but… well, as I said elsewhere:
I think that makes familiarising yourself with those arguments (whether from the debate or another equivalent-or-better source) a prerequisite for making the kind of strong, confident claims Roko is making. At the moment, he’s making those claims without the information necessary to be anywhere near as confident as he is.
There’s a good reason why the foundational text in an academic field would be a text and not a video. Evaluating arguments is easier to do when they are done via text.
I don’t think anyone has shown that the debate contains specific arguments that Roko is unfamiliar with.
See: https://www.lesswrong.com/posts/ZdNyKE5yC4YjXGzfG/a-back-of-the-envelope-calculation-on-how-unlikely-the?commentId=cWCMuMpMQ98gqRB2G
Which to me strongly suggest Roko was unfamiliar with multiple (imo) strong evidence for the zoonosis origin.
Isn’t there a transcript? In any case, this seems to be highly subjective, and in my opinion not hugely relevant anyway. To extend the analogy, your expectation of someone’s having read foundational texts before making strong claims would hardly be lessened by the objection that the texts were hard to read.
Do we need to positively show that? As I mentioned, many intelligent, thoughtful people who were already very familiar with this question and its relevant facts updated significantly based on the debate. And Roko certainly hasn’t explicitly addressed all of the arguments therein. Isn’t that enough to suggest that he should at least watch it?
It’s equivalent to watching a season of a TV show- is that really such an onerous requirement for making incredibly strong, potentially libellous claims about a contentious issue with serious real-world ramifications?
No transcript, but the judges have documents where they outline their reasoning:
Will van Treuren’s doc
Eric Stansifer’s doc
If you watch a debate the way you would a TV show where you aren’t critically evaluating everything, that’s not a good basis for forming beliefs about the real world.
Youtube videos tend to encourage you to consume them in that way, but that doesn’t make it a good epistemic practice and that’s part of the reason why scientific debates usually happen in text.
I watch TV in a pretty focused way where I take things in.
But I wasn’t suggesting you watch it like a TV show; just that’s a similar time commitment (ie not an unreasonable one).
Notes:
Yep, the tweet thread is just me jotting down thoughts etc while watching the debate.
I was 75-80% convinced when I tweeted that, which was before I had finished the debate. After watching the debate, I made a sketchy Bayesian calculation that got me to 96%, but I’ve since backed off to maybe 66%.
Basically: the key question for me is whether you think one of the first outbreaks of COVID happened at the Huanan Seafood Market. Rootclaim conceded this, and as far as I can tell if this is true then it’s dispositive evidence, but I have since began to doubt it.
One thing in favour of my judgements is that I’m at number 5 on Metaculus’ leaderboard of how accurate predictors were compared to their peers, altho that mostly comes from predictions made between 2016 and April 2020, when I burnt out from forecasting on Metaculus.
I’m currently in the Diamond league on Manifold, which is much less impressive, and in part driven by my prediction of which way the debate would go.
Oh—it could be that my peer score is artificially high due to using Metaculus back when there were fewer peers who were good at forecasting.
Sure, my post doesn’t rebut Roko’s. It isn’t an arguement, after all. It is a recommendation for some content.
Yep, I didn’t watch the whole debate. I apologize if I gave that impression. Mea culpa.
How much you update on the evidence depends on your priors, no? Daniel went from “probably lab-leak” to “probably zoonosis”, and I went from “IDK” to “quite probably lab-leak”. That is the similairity I was pointint at.
I found Daniel’s thread helpful after watching some hours of the debate, as he noticed numerous points I missed. Plus, I had watched enough that the rest of the data in the thread made sense to me. So you’re right, Daniel’s thread probably isn’t that helpful. At least, not without watching e.g. the opening arguements by rootclaim and peter miller.