This is indeed quite well-written and a helpful summary! One question I have: You write that “Fortunately, Prof. Bostrom’s later comments suggest that he, at least, no longer believes in the first claim to the full extent he once did.” Where are you getting that?
The third claim is indeed what has attracted the most attention and censure, but it’s the interaction with the first that makes it particularly toxic by ruling out many of the more charitable readings. It’s not just that he thinks the sentence “Blacks are more stupid than whites” is factually true but regrettable; he wrote that he “likes” that sentence, particularly because “the more counterintuitive and repugnant a formulation, the more it appeals to me given that it is logically correct.”
It would be therefore great if Professor Bostrom explained somewhere why he no longer personally finds repugnant formulations of true statements appealing, but I couldn’t find it in the apology linked above.
Eyeballing the chart, it seems like more than a “years since pledge” dependence, there’s a “actual year” dependence independent of the pledge year, at least after year 0. The most stark change is from 2010 to 2011, when reporting rates increased significantly for both of the first two cohorts. I’m guessing that this is due to some technicality rather than an actual behavior change. There also seems to be a significant decay across all cohorts from 2015 to 2016 that would likely be even more obvious if plotted. Maybe this also has a boring explanation, or maybe it has to do with identifiable events in 2016.