It's hard to describe 16 different types. MBTT is already criticised because it's pretty hard to know the behaviours of 16 separate types. It's harder than 4 temperaments, 4 main DISC profiles, or 5 FFM dichotomies, or even 9 enneagram types.
Myersian types are often badly described. For example, I don't like descriptions of my own myersian type, which say that ENTJ's are pushy assholes or such (i.e. describing ENTJ's as -++, as they can be 0+- as well).
Socionic types are badly described too, but in a lesser extent. For example, are SEI's accomodating ? EIE's sociable ? SLE's pushy ? SLI's sluggish ? LII's quiet ? LIE's outgoing ?
It depends of the physiological variant. For example, a --- SEI may appear as SLE to an unexperienced socionist. Or, a 0+- LII as ILE. Another example is that even experienced socionists tend to write S0D0 like descriptions of LSI's, while putting lots of S-D+ people on LSI celebrity lists, such as Stalin or Milosevic.
Now that there is GCPT, we have a set of 168 different types. It's already hard to deal with 16 types, it's harder to deal with 168.
Is someone interested about writing such descriptions ? Providing examples ? It'd prevent those who get to know socionics to deal with stereotypes, instead of types.