If instead of each person checking if an opinion is held by the majority of the people they know,
and seeing what the typical result for that would be,
what if the operation was doubled?
So, each person would check what the majority opinion is among those they know, and then store that value.
Then each person would check what the majority of /those/ values are among the people they know.
I was wondering if this might partially undo the effects from some people being known to more people.
so, if a person x 's opinion on a topic is f(x), and the friends of a person is x.friends
then instead of majority(map(f,x.friends)) , instead use majority(map(lambda z:majority(map(f,z.friends)), x.friends))
So instead of using "what opinions do I see", instead using "what opinions do I see people saying they see". Would that be a more or less accurate measure of "what opinions do people have"?
