I know that one :-) I especially liked the part where the student discovered that the predictions were met better than 'possible' (probably by some convergence speed theorems).

But no. I thought more about something along the lines of http://en.wikipedia.org/wiki/Behavioral_Economics (actually mentioned on your link).

Some theory which doesn't try to predict aggregate outcomes of human actions (that wouldn't work for longer time spans for the same reasons whether cannot be forcast precisely - chaos theory). I thought about some theory which models smaller structures like transactions, rituals, customs, ceremonies and measures and predicts their frequency and 'success'.