The value statement in the article appears to be; "make predictions, assess the quality of those predictions, adjust your prediction engine, repeat"
I think that this blog often falls into the trap of too much jargon without enough context, because it uses large chunks of direct quotes.
I don't think that it over reached with reference class forecasting - it seemed to me to do a good job of summarazing. Which step do you feel overreached?
I don't mean to be disrespectful to you, since you've asked a good question, but I didn't value reading it enough to go back and identify which parts over reached. If you a pressing enough reason, I can reconsider. Alternatively, just check out some of the original papers and see what you think yourself. Edit 1: I might add, that there are alternative ways of taking the outside view. Amongst them is setting accuracy oriented goals to counter emotionally motivated reasoning
Great article that systemetizes the process of taking the outside view into 4 steps. Thought LWers might get a kick out of it.
http://www.farnamstreetblog.com/2015/05/inside-view-michael-mauboussin/