As far as I can tell, Kurzweil's methodology explicitly predicts a vaguely defined "rate of change" that is currently roughly doubling every fifteen years. Within that methodology, he extracts fairly strait-forward preditions, assumes instant adoption of technologies once they are feasible and ignores black swans, and largely ignores basic science in general focusing on technology. In addition he adds a few idiosyncratic predictions that match his interests and which are not calibrated to the general rate of change being predicted. In particlar, these predictions tend to accurately estimate near future computer pattern recognition capabilities in natural domains but grossly underestimate human pattern recognition in such domains and to be grossly overoptimistic about market enthusiasm for computer systems which underperform relative to humans in pattern recognition functions. The predictions on the above list generally seem to fit this methodology.