As a metaphore for understanding the people prefering a "change of
paradigm" definition of complexity, consider Persig's book ZEN AND
THE ART OF MOTORCYCLE MAINTENANCE. In his book, Persig talked of
"quality" and how it couldn't be defined, much less measured in any
conventional way. Perhaps "complexity", like "quality", is of
such a nature that it is illusive to conventional methods...perhaps
requiring a different way of looking at it...perhaps it is
necessary not to look at it as an object (something you can attach
a number to)...but as something that can only be approached by
looking at the relationships or organization of certain interacting
components...and perhaps the distinguishing thing separating the
"complex" from the "non-complex" is the existance of a closed
causal loop, self-reference, strange loop, inpredicativity, or
vicious circle (all terms for the same thing). This is what Rosen
claims. Also, interestingly, Onar pointed out a while back that
hyper-set theory may be a way to resolve such paradoxes...although
I haven't had time to look into this personally.
Jeff Prideaux