So I’d give Kurzweil 10-20% here, not 50+%. I think the main reason is that SA was prepared to give Kurzweil a “weakly true” if most of a prediction was solid, whereas I required every part of a prediction to be basically right. If I broke the predictions down into individual sentences and scored those sentences one by one, Kurzweil would score higher.
I’m very much of mind that scoring this way was the correct course, since the predictions are cunjunctive, and for any prediction to be strong evidence for his overarching claims all its parts must be coherent and true. Intuitively, applying Occam’s Razor should, IMO, produce the results you’ve obtained.
I’m very much of mind that scoring this way was the correct course, since the predictions are cunjunctive, and for any prediction to be strong evidence for his overarching claims all its parts must be coherent and true. Intuitively, applying Occam’s Razor should, IMO, produce the results you’ve obtained.