our sessions and guests were mostly not controversial — despite what you may have heard
Fair, although it is also fair to characterize the controversial content as fairly substantial and not isolated. Looking at RSVPs on the Saturday schedule for content appears to be controversial:
Collins and Collins had 57 RSVPs, second in the timeslot
[For purposes of this exercise, I did not code Robin Hanson or Scott Alexander as “controversial,” in part because their sessions seemed to clearly focus on non-controversial content.]
In my view, a neutral observer who reviewed the schedule and RSVP numbers would likely conclude that eugenics was a major topic/focus of the conference relative to most topics (but behind forecasting/prediction). The platforming of people with problematic views on race should be considered in that context.
I haven’t found much information about Smith, so the coding of that session as controversial is tentative. I preliminarily coded that way because the post I found clearly advocated for practical eugenics, and so at least thematically fit in with the rest of the “eugenics / eugenics-adjacent track.”
For context, the total number of RSVPs on the conference schedule was around 14,000 (edit: this is for the whole conference season, for Manifest in-particular it was closer to 7000). So by “substantial” you mean something like 279⁄14,000 = ~2% (edit: or 4% limiting it to just Manifest).
Given that the quote was “our,” was written by Austin, and was in a post about Manifest, I don’t see the argument for including non-Manifest RSVPs in the denominator. Indeed, my sample was events on Saturday (the day’s schedule that Austin screenshotted in the post), although it does appear that the eugenics content was concentrated on that day,
I would probably exclude RSVPs to things like “dinner,” “party,” social events, etc. as non-substantive. I also have no way to discern how much “scientific” racism was going on at dinner or parties. Removing these from the denominator implicitly sets their racism percentage as equal to the racism percentage of the substantive sessions, which seems more likely than assuming they were at 0%. Stated differently, dinner is just dinner; it does not dilute the relative concentration of any particular substantive content.
Admittedly, the percentage of these RSVPs over the entire adjusted Manifest event RSVPs as denominator will presumably be in the single digits. Is that “fairly significant” (I used the qualifier in my comment)? I would say yes:
“Fairly significant” is a flexible term, and I think the necessary quantum is ordinarily lower when we’re talking about objectionable content. To take an extreme example, suppose you don’t like your child hearing a certain word that appears 50 times in a two-hour movie. That’s only ~50 seconds out of 7200, or 0.694 percent. But most parents would call far less than 50 uses of a particular word in a movie fairly significant! I don’t mean to suggest that 0.694 percent is per se significant, only that this kind of percentage analysis can miss the mark when objectionable content is involved.
I expect that conducting a similar percentage analysis on other topics would show the eugenics/controversial cluster fairly high on the list of clusters. This is a form of relative significance; the topic drew a relatively high amount of interest when compared to various other topics at hand. Moreover, in the examples I mentioned, the talk was first or second out of perhaps five or six options. Sometimes the most-RSVP’d event was something expected to be very popular (like a Robin Hanson talk), and the other options were generally not weak. I think that provides some support for a conclusion that the content was fairly significant in context.
I expect that conducting a similar percentage analysis on other topics would show the eugenics/controversial cluster fairly high on the list of clusters. This is a form of relative significance; the topic drew a relatively high amount of interest when compared to various other topics at hand.
I think I would take that bet, in the sense that if you actually ranked all clusters (including overlapping clusters) of content at this level of coherence of a cluster, that it probably wouldn’t rank in the top 10. This is of course not a great operationalization, and it would take a lot of work to get to an actual bet, but as someone who attended the event, my current best guess is we disagree on a relatively concrete object-level claim.
Given that the quote was “our,” was written by Austin, and was in a post about Manifest, I don’t see the argument for including non-Manifest RSVPs in the denominator.
Agree! I just accidentally looked up the wrong number in my comment, and then didn’t want to stealth-edit it.
Fair, although it is also fair to characterize the controversial content as fairly substantial and not isolated. Looking at RSVPs on the Saturday schedule for content appears to be controversial:
Collins and Collins had 57 RSVPs, second in the timeslot
Anomaly had 60 RSVPs, second in the timeslot
G. Smith had 86 RSVPs, first in the timeslot[1]
Khan had 76 RSVPs, second in the timeslot
[For purposes of this exercise, I did not code Robin Hanson or Scott Alexander as “controversial,” in part because their sessions seemed to clearly focus on non-controversial content.]
In my view, a neutral observer who reviewed the schedule and RSVP numbers would likely conclude that eugenics was a major topic/focus of the conference relative to most topics (but behind forecasting/prediction). The platforming of people with problematic views on race should be considered in that context.
I haven’t found much information about Smith, so the coding of that session as controversial is tentative. I preliminarily coded that way because the post I found clearly advocated for practical eugenics, and so at least thematically fit in with the rest of the “eugenics / eugenics-adjacent track.”
Some further data: even though I took part in some of these sessions, I didn’t RSVP to them since RSVP status was public through the schedule website.
For context, the total number of RSVPs on the conference schedule was around 14,000 (edit: this is for the whole conference season, for Manifest in-particular it was closer to 7000). So by “substantial” you mean something like 279⁄14,000 = ~2% (edit: or 4% limiting it to just Manifest).
Given that the quote was “our,” was written by Austin, and was in a post about Manifest, I don’t see the argument for including non-Manifest RSVPs in the denominator. Indeed, my sample was events on Saturday (the day’s schedule that Austin screenshotted in the post), although it does appear that the eugenics content was concentrated on that day,
I would probably exclude RSVPs to things like “dinner,” “party,” social events, etc. as non-substantive. I also have no way to discern how much “scientific” racism was going on at dinner or parties. Removing these from the denominator implicitly sets their racism percentage as equal to the racism percentage of the substantive sessions, which seems more likely than assuming they were at 0%. Stated differently, dinner is just dinner; it does not dilute the relative concentration of any particular substantive content.
Admittedly, the percentage of these RSVPs over the entire adjusted Manifest event RSVPs as denominator will presumably be in the single digits. Is that “fairly significant” (I used the qualifier in my comment)? I would say yes:
“Fairly significant” is a flexible term, and I think the necessary quantum is ordinarily lower when we’re talking about objectionable content. To take an extreme example, suppose you don’t like your child hearing a certain word that appears 50 times in a two-hour movie. That’s only ~50 seconds out of 7200, or 0.694 percent. But most parents would call far less than 50 uses of a particular word in a movie fairly significant! I don’t mean to suggest that 0.694 percent is per se significant, only that this kind of percentage analysis can miss the mark when objectionable content is involved.
I expect that conducting a similar percentage analysis on other topics would show the eugenics/controversial cluster fairly high on the list of clusters. This is a form of relative significance; the topic drew a relatively high amount of interest when compared to various other topics at hand. Moreover, in the examples I mentioned, the talk was first or second out of perhaps five or six options. Sometimes the most-RSVP’d event was something expected to be very popular (like a Robin Hanson talk), and the other options were generally not weak. I think that provides some support for a conclusion that the content was fairly significant in context.
I think I would take that bet, in the sense that if you actually ranked all clusters (including overlapping clusters) of content at this level of coherence of a cluster, that it probably wouldn’t rank in the top 10. This is of course not a great operationalization, and it would take a lot of work to get to an actual bet, but as someone who attended the event, my current best guess is we disagree on a relatively concrete object-level claim.
Agree! I just accidentally looked up the wrong number in my comment, and then didn’t want to stealth-edit it.