Strongly agree with your comment. I noticed the same thing while reading the post. The factor i for average impact per participant is really handwavy; I think impact per person varies widely (even within preselected groups, e.g. 80k’s target audiences such as Numberphile youtube viewers) and this makes the model less useful at this level of simplification.
Strongly agree with your comment. I noticed the same thing while reading the post. The factor i for average impact per participant is really handwavy; I think impact per person varies widely (even within preselected groups, e.g. 80k’s target audiences such as Numberphile youtube viewers) and this makes the model less useful at this level of simplification.