simplifying in science is a regular ocfurance, especially physcis, we study the most simple system first - reductionism to reduce our system to the simplest parts, and attempt to study them seperately...
however physics has reached the limits of reductionism. We understand that complex behaviour (like in foams, stockmarkets, betting website, etc.) can not be understand be just looking at the simplest components.
We understand the concept of an emergent property, and how to describe it without reference to the underlying components.
and thus we've found the limitations of 'simplification' maybe other scientists haven't but that's not to say they are all doing a crap job.
Your point is that humans are simple, and i vhemently dispute that claim. BMy point is you can't do an experiment on an entire economy, and the emergent complex features are what throw all your simple models off.
The kinds of experiment you are talking about are uncontrolled historic examples. The error in a count is the square root of the count, thus if you can count 1 example the error is +/-1 (which gives arange of 0-2, or the possiblility that ever single example was just a fluke). I'm pretty sure no examples are precisely the same...
i've yet to see a good book, but i'd love to read this one and then continue my rant.